VIS · GPT-5.6-Terra · C6+

Confirmed. I have absorbed the workflow library and its range of approaches, including temporal/event analysis, movement and network workflows, cohort comparison, visual machine learning, clustering, simulation steering, text analysis, and multimodal interaction analysis.

I will use the library as a design resource and cite relevant workflows by name when proposing the workflow for your task. I have no questions at this stage.

Evaluating citation networks

I'm considering how to handle a citation network if available, looking at time distributions and topic drift. I really need to avoid assigning the research to a multi-community paper. Instead, I'll assign the paper's contributions based on the authors’ community memberships and their respective fractions. I wonder how the leaders within those communities can influence this as well. It's definitely a bit of a balancing act!

Recommended workflow: Dynamic, overlapping co-authorship communities with topical characterization

This workflow treats the VIS literature as an evolving collaboration network, while preserving the fact that researchers may belong to several communities. It combines ideas from NodeTrix, Reducing Snapshots to Points, Human-Steered Topic Modelling, SOMFlow, and StratomeX.

1. Audit and prepare the publication data

First, validate the paper–author records:

• Use DBLP author identifiers as the primary identity key; retain raw names for display and manual checking.
• Identify duplicate papers, incomplete author lists, missing abstracts/keywords, and unusually large author teams.
• Inspect yearly publication counts by venue, because the four venues have different histories and may create apparent community changes caused by coverage rather than genuine intellectual change.
• Normalize titles, abstracts, and keywords: remove generic terms such as visualization, visual analytics, and venue names; retain domain terms and methodological phrases.
• If “citations” means citation links between papers, construct a paper-citation network. If it means only citation counts, use those as impact indicators rather than network links.

Create two linked data structures:

1. A paper–author bipartite network, preserving full author teams.
2. A weighted co-authorship network, where authors are nodes and a collaboration edge is created for each co-authored paper.

To avoid large author teams dominating the network, fractionally weight each paper’s collaboration contribution. For a paper with (n) authors, distribute a total contribution of 1 across its author pairs rather than assigning every pair a full weight.

2. Define temporal views at more than one scale

Construct overlapping temporal snapshots, for example:

• annual snapshots for detailed inspection;
• rolling 3- or 5-year windows for more stable community detection;
• broader eras, such as 1990–1999, 2000–2009, 2010–2019, and 2020–2024, for high-level reporting.

The analyst should interactively vary the window width. Short windows reveal rapid membership changes but may fragment communities; longer windows reveal durable structures but can conceal splits or temporary collaborations.

For each snapshot, calculate:

• weighted co-authorship strength;
• publication count;
• weighted degree and within-community degree;
• betweenness or brokerage measures;
• number of recurring collaborators;
• venue composition;
• citation impact, where available.

3. Discover candidate overlapping communities

Use an overlapping community model, not a hard partition. Suitable approaches include link-community detection, mixed-membership stochastic block models, or overlapping modularity methods. The important design choice is that each author receives a membership weight for one or more communities in each time window.

Run the detection at several resolutions and parameter settings. Do not treat one algorithmic result as definitive. Instead, compare candidate solutions according to:

• number and size distribution of communities;
• internal collaboration density;
• stability under bootstrap resampling of papers or edges;
• persistence of communities across adjacent time windows;
• interpretability of their topical profiles;
• plausibility according to visualization-domain knowledge.

The analyst should use a community-resolution control to merge overly fragmented clusters or examine subcommunities within large ones. This follows the human-guided refinement logic of SOMFlow: Guided Exploratory Cluster Analysis.

Report community number in two forms:

• the number of active communities per time window;
• the number of distinct community lineages across the full 1990–2024 period.

This distinction prevents temporary splits or renamed continuations from being counted as entirely new communities.

4. Identify cores, leaders, and peripheral members

For every community and time window, classify members by both structural role and temporal persistence.

A practical classification is:

• Stable core: authors with sustained, high membership in the community over multiple adjacent windows and repeated collaborations with other core members.
• Leaders: stable-core authors with high within-community collaboration strength, high publication contribution, or strong brokerage between the community and others.
• Peripheral contributors: authors with limited, short-lived, or low-strength membership.
• Boundary spanners: authors with meaningful membership in multiple communities or high cross-community collaboration.

Do not define leaders solely by publication count or citation count. Present a composite profile including:

• years active in the community;
• within-community weighted degree;
• number of repeated co-authors;
• cross-community brokerage;
• paper and citation impact;
• topical contribution.

The analyst should be able to inspect the underlying paper list for any proposed leader, especially when a high centrality score is driven by one large team paper or a short-lived collaboration.

5. Track community evolution and lineage events

Match communities between consecutive temporal snapshots using a combination of:

• weighted author-membership overlap;
• overlap of stable-core members;
• topical similarity;
• collaboration-structure similarity.

Use core overlap more heavily than peripheral overlap when deciding whether a community persists. A community whose peripheral membership changes substantially but whose core remains stable should normally be treated as a continuation.

Classify temporal events using explicit rules, subject to analyst review:

Visualize the result as a community lineage diagram: communities appear as bands over time, band width encodes active membership or publication volume, and links encode continuation, splitting, and merging. Membership overlap should be visible separately from lineage; otherwise, a few mobile authors can falsely imply a merger.

Use the dynamic-network perspective of Reducing Snapshots to Points as a complementary overview. Represent each temporal snapshot by a vector of network and community features—such as community-size distribution, core persistence, overlap level, modularity, and topic mixture—and project snapshots into a “network-state map.” This can reveal periods with unusually rapid restructuring, such as the emergence of new venues, interdisciplinary expansion, or changes in collaboration scale.

6. Characterize each community’s research foci

For every community–time-window pair, build a weighted document corpus from papers authored by its members.

When an author belongs to multiple communities, fractionally allocate the paper to each community according to that author’s membership weight. Retain a second view of full paper membership so the analyst can inspect interdisciplinary papers without forcing them into one topic.

Apply an initial topic model, preferably NMF or another interpretable topic method, to titles, abstracts, and keywords. Then use the interactive refinement strategy from Human-Steered Topic Modelling:

• merge redundant topics;
• split topics that mix distinct research areas;
• remove generic terms;
• seed missing domain concepts using key terms or selected papers;
• compare topic assignments against paper-level evidence.

For each community, show:

• its top topics and representative keywords;
• representative papers for each topic;
• topic prevalence over time;
• emerging, declining, and persistent topics;
• topic diversity and topical concentration;
• topics shared with other communities.

If citation links are available, add citation-flow summaries: which communities cite one another, which external fields they draw from, and whether a topical change coincides with new citation connections.

7. Provide coordinated visual analysis views

A research analyst should work across linked views rather than rely on a single network picture.

For overlap analysis, the analyst should be able to select two or more communities and see:

• shared authors;
• authors with strong membership in both;
• shared papers;
• common topics;
• cross-community collaboration edges;
• whether overlap is stable, increasing, or limited to a particular period.

This follows the set-comparison reasoning used in StratomeX, adapted from patient-set overlap to author-community overlap.

8. Investigate major evolutionary events with evidence

When the lineage view identifies a split, merge, reorientation, or dissolution, open an event-focused comparison workspace:

1. Compare the predecessor and successor communities’ stable cores.
2. Identify members who stayed, left, joined, or became boundary spanners.
3. Compare their collaboration matrices before and after the event.
4. Compare topic distributions, representative papers, and key terms.
5. Inspect venue shifts and, if available, citation flows.
6. Review a timeline of highly central papers and new collaborations around the event.
7. Record an analyst interpretation of the likely cause.

Potential interpretations should be framed as evidence-backed hypotheses, not automatic causal claims. For example:

• a merge may coincide with increased collaboration between previously separate methodological and application-oriented groups;
• a split may reflect divergence into visual analytics versus domain-specific visualization work;
• a topical reorientation may coincide with new data types, algorithms, venues, or influential papers;
• a dissolution may be an artifact of sparse publication periods rather than the end of an intellectual community.

9. Iterate, validate, and document uncertainty

The analyst should repeat detection and tracking under reasonable alternatives:

• different rolling-window widths;
• different edge-weighting schemes;
• different community resolutions;
• inclusion versus exclusion of weak, one-off ties;
• alternative topic-model sizes and preprocessing choices.

Flag conclusions that are sensitive to these choices. In particular, distinguish:

• robust communities, visible across parameter settings;
• tentative communities, present only at a narrow resolution or in sparse periods;
• uncertain lineage events, where several predecessors or successors are similarly plausible.

Maintain an analysis history containing parameters, manually adjusted community boundaries, topic edits, event labels, and supporting paper-level evidence. This makes the resulting intellectual-history account reproducible and open to later revision.

Expected analytical outcomes

The workflow should yield:

1. A defensible estimate of the number of active communities and long-term community lineages.
2. For each community, a profile of its stable core, leaders, peripheral members, and boundary spanners.
3. A temporally resolved account of membership overlap and researcher mobility.
4. Topic profiles showing each community’s major research foci and topical evolution.
5. A reviewed event history of community births, merges, splits, reorientations, and dissolutions.
6. Evidence-backed interpretations linking structural changes to papers, collaborations, topics, venues, and citation patterns.

Listing relevant approaches

I need to compile a list of relevant papers from the response, and it’s important to be honest and concise. The response mentioned NodeTrix, Reducing Snapshots, Human-Steered, SOMFlow, and StratomeX. I should focus on those five, as the workflow explicitly combines ideas from them. Additionally, I need to ensure that I cite these within the text appropriately, sticking to the request for English and highlighting their relevance.

Relevant library papers and the approaches used:

1. Reducing Snapshots to Points: A Visual Analytics Approach to Dynamic Network Exploration
   Stef van den Elzen et al., 2016
   - Used for representing time-windowed co-authorship networks as evolving network states.
   - Informs the snapshot-state projection for finding stable periods, transitions, and unusual restructuring.

2. NodeTrix: A Hybrid Visualization of Social Networks
   Nathalie Henry et al., 2007
   - Used for detailed inspection of dense collaboration communities while preserving links between communities.
   - Informs the proposed matrix-plus-node-link view for core structure, internal collaboration, and boundary-spanning authors.

3. Human-Steered Topic Modelling / UTOPIAN: User-Driven Topic Modeling Based on Interactive Nonnegative Matrix Factorization
   J. Choo et al., 2013
   - Used for characterizing communities’ research foci from titles, abstracts, and keywords.
   - Informs analyst-guided topic refinement: merging, splitting, seeding, and editing topics rather than accepting an automatic topic model uncritically.

4. SOMFlow: Guided Exploratory Cluster Analysis with Self-Organizing Maps and Analytic Provenance
   D. Sacha et al., 2018
   - Used for the iterative, human-guided approach to community detection and resolution selection.
   - Informs maintaining provenance of parameter choices, community refinements, and interpretations across analysis iterations.

5. Guided Visual Exploration of Genomic Stratifications in Cancer (StratomeX)
   M. Streit et al., 2014
   - Used for comparing overlapping groups and tracing membership across alternative or evolving groupings.
   - Informs the community-overlap views, membership comparisons, and linked inspection of shared members and associated characteristics.

6. Spatio-temporal Analysis and Modelling
   N. Andrienko and G. Andrienko, 2013
   - Relevant to iterative clustering of temporal profiles and analyst validation of cluster coherence.
   - Informs the recommendation to inspect and refine temporal community groupings rather than treat clustering as final.

7. Visual Analysis of Multivariate State Transition Graphs
   A. J. Pretorius and J. J. van Wijk, 2006
   - Relevant to analyzing transitions among abstract states.
   - Informs the explicit treatment of community continuations, splits, merges, births, and dissolutions as interpretable evolution events.

8. Jigsaw: Supporting Investigative Analysis through Interactive Visualization
   J. Stasko et al., 2008
   - Relevant to grounding high-level patterns in original textual evidence.
   - Informs the paper-level inspector, where analysts validate community-topic interpretations using underlying titles, abstracts, authorship records, and citations.

No questions. The ATWL specification and example are clear. I will represent the proposed workflow using explicit artifacts, transforms, analyst/machine roles, iterative refinement loops, assessment outcomes, and specifications where they were implied by the original design.

Considering topic refinement

I'm thinking about creating a more effective "robustness and refinement loop" to enhance snapshot rebuilding. There seems to be this idea of nested loops, which might complicate things. My natural language processing involves iterative detection and topic model refinement, including reviewing events and tracking alternatives. I'm considering if we can combine all these into one coherent loop. I envision a structured flow: initial construction, community discovery, topic refinement, visualization, and validation. It’s really about linking topics with final candidate communities.

Refining tracking parameters

I'm thinking about tracking alternate parameters and refining topics. I wonder if nested loops are allowed but should probably avoid them. Maybe I can have one main loop for community configuration and then create sub-loops for topics and lineages. However, to update the community configuration, I’d need to return to an earlier state, which could complicate control.

It seems that validation doesn't have to change the results but can track options, focusing on sensitivity using alternative machine assessments. I should consider initial auditing to evaluate quality and refine cleaning processes.

Validating and refining processes

I’m considering the validation process, which suggests I first inspect before looping. Instead of creating fabricated loops, I could define a hybrid unit for visualization. Injecting domain knowledge, like K_domain, will help with interpretations and resolutions.

I should also focus on how to declare artifacts correctly before referencing them. For example, my assignments should clearly state identifiers and ensure everything is succinct yet sufficient. If I’m preparing data, I must include quality audits and, if needed, assess human input. The original document didn’t specify a correction loop, which is interesting!

Processing data and author profiles

I’m noting that there are imperfect data flags that go into filtering and normalization processes. When preparing data with T_prepare, I can produce clean outputs. For authors, I’m considering D_authors, which represent a group aggregated by them. Coauthorship snapshots can be time-partitioned for aggregation.

I might characterize clean papers using text features like F_text. For author profiles, the T_author_profiles will provide meaningful metrics based on communities. Finally, I should derive network snapshots, defining them as D_snapshots, all while ensuring that the original two linked structures are preserved.

Clarifying ATWL representation

It looks like the user wants to focus on the ATWL representation, so I should ensure that I include detailed outputs. I’ll skip some feature declarations to keep it concise but make sure the code covers all output declarations after the primary outputs. I also need to consider the origin semantics in my explanation and jot down brief notes. Keeping everything organized is essential for clarity!

workflow vis-community-evolution
  template: characterise -> visualise -> assess -> define-unit ->
            characterise -> define-unit ->
            loop(build-model -> define-unit -> characterise ->
                 contextualise -> visualise -> abstract -> assess) ->
            characterise ->
            loop(build-model -> define-unit -> characterise ->
                 visualise -> abstract -> assess) ->
            define-unit -> characterise -> contextualise -> visualise ->
            abstract -> generate-knowledge ->
            loop(build-model -> define-unit -> characterise ->
                 visualise -> assess) ->
            generate-knowledge
  description: "Identify overlapping, evolving co-authorship communities
    in IEEE VIS publications; characterise their stable cores, leaders,
    research foci, membership overlap, and community lineage events."

# ============================================================================
# INPUT DATA, DOMAIN KNOWLEDGE, AND INITIAL SETTINGS
# ============================================================================

artifact D_papers_raw : entities
  origin: given
  internal structure: elementary
  embedment: {set, time}
  features:
    - id: paper_id
      value structure: atomic
      value type: reference
      description: "Unique paper identifier"
    - id: publication_year
      value structure: atomic
      value type: temporal
      description: "Publication year from 1990 to 2024"
    - id: venue
      value structure: atomic
      value type: categorical
      description: "VIS, InfoVis, VAST, or SciVis venue"
    - id: authors
      value structure: list
      value type: reference
      description: "DBLP-disambiguated author identifiers and raw names"
    - id: bibliographic_text
      value structure: vector
      value type: text
      description: "Title, abstract, and author keywords"
    - id: citations
      value structure: relational configuration
      value type: reference
      description: "Citation links where available, otherwise citation counts"
  description: "IEEE VIS conference paper records with authorship,
    bibliographic text, venue, year, and citation information"

artifact K_domain_knowledge : knowledge(D_papers_raw)
  origin: given
  representation form: "annotations and interpretive criteria"
  description: "Visualization-domain knowledge used to judge meaningful
    communities, distinguish venue-history effects from intellectual
    change, interpret topics, and review lineage events"

artifact S_temporal_views : specification
  origin: given
  representation form: "configuration schema"
  description: "Annual, rolling three- or five-year, and broader-era
    temporal views; specifies the candidate temporal windows to retain"

artifact S_community_analysis : specification
  origin: given
  representation form: "configuration schema"
  description: "Initial community-analysis choices: fractional weighting
    of co-authorship ties, overlapping community method, candidate
    resolutions, weak-tie treatment, and selection criteria"

artifact S_topic_analysis : specification
  origin: given
  representation form: "configuration schema"
  description: "Initial topic-analysis choices: text preprocessing,
    generic-term removal, topic-model family, number of topics, and
    topic-editing constraints"

artifact S_lineage_matching : specification
  origin: given
  representation form: "rules and thresholds"
  description: "Rules for matching communities across periods using
    weighted author membership, stable-core overlap, topical similarity,
    and collaboration-structure similarity"

artifact S_robustness_analysis : specification
  origin: given
  representation form: "parameter settings"
  description: "Alternative window widths, edge weights, weak-tie filters,
    community resolutions, and topic-model settings for sensitivity checks"

# ============================================================================
# DATA AUDIT AND PREPARATION
# ============================================================================

transform T_audit_records :
  intent: characterise
  manner: "compute data-completeness and coverage statistics"
  input: D_papers_raw
  output: F_record_quality
  actor: machine
  description: "Identify duplicate papers, missing author or text fields,
    unusually large author teams, yearly venue coverage, and the available
    form of citation information"

artifact F_record_quality : feature(D_papers_raw)
  value structure: vector
  value type: {numeric, categorical}
  description: "Completeness, duplication, author-team-size, venue-coverage,
    and citation-availability indicators for paper records"

transform T_visualise_audit :
  intent: visualise
  manner: "temporal coverage and quality overview"
  input: D_papers_raw, F_record_quality
  output: V_data_audit
  actor: machine
  description: "Show yearly paper counts by venue and linked views of
    missingness, duplicates, author-team sizes, and citation coverage"

artifact V_data_audit : visualisation(D_papers_raw, F_record_quality)
  layout: "aligned yearly timelines and linked quality summaries"
  form: "stacked bars, distributions, and tabular record inspector"
  encoding: "time and venue determine publication-count displays; colour
    and highlighting indicate data-quality conditions"
  description: "Visual audit interface for checking data coverage and
    identifying records requiring treatment"

transform T_assess_audit :
  intent: assess
  manner: "judge data adequacy and required cleaning"
  input: V_data_audit, F_record_quality, K_domain_knowledge
  output: K_audit_assessment
  actor: human
  description: "Determine how duplicate records, missing information,
    unusual team sizes, and venue-coverage changes should be handled"

artifact K_audit_assessment : knowledge(D_papers_raw, F_record_quality)
  representation form: "quality judgment and cleaning decisions"
  description: "Analyst assessment of data quality and justified treatment
    decisions for the bibliographic corpus"

transform T_prepare_records :
  intent: define-unit
  manner: "filter, deduplicate, and normalise paper records"
  input: D_papers_raw, K_audit_assessment
  output: D_papers
  actor: hybrid
  description: "Create an analysis-ready paper corpus; retain DBLP author
    identifiers, preserve display names, normalise bibliographic text, and
    apply agreed data-treatment decisions"

artifact D_papers : entities
  internal structure: elementary
  embedment: {set, time}
  features:
    - id: publication_year
      value structure: atomic
      value type: temporal
      description: "Validated publication year"
    - id: venue
      value structure: atomic
      value type: categorical
      description: "Publication venue"
    - id: author_ids
      value structure: list
      value type: reference
      description: "Validated DBLP author identifiers"
    - id: clean_text
      value structure: vector
      value type: text
      description: "Normalised title, abstract, and keywords"
  description: "Prepared IEEE VIS paper corpus for collaboration and topic analysis"

# ============================================================================
# AUTHOR, TEXT, AND TEMPORAL NETWORK REPRESENTATIONS
# ============================================================================

transform T_define_authors :
  intent: define-unit
  manner: "group-by-entity"
  input: D_papers
  output: D_authors, F_authorship
  actor: machine
  description: "Define unique researchers from DBLP identifiers and retain
    paper-to-author affiliations"

artifact D_authors : entities
  internal structure: group/cluster
  embedment: set
  features:
    - id: author_id
      value structure: atomic
      value type: reference
      description: "Unique DBLP author identifier"
    - id: display_names
      value structure: list
      value type: text
      description: "Raw author-name strings retained for inspection"
  description: "Researchers represented through their associated VIS papers"

artifact F_authorship : feature(D_papers)
  value structure: relational configuration
  value type: reference
  representation form: "paper-author bipartite network"
  description: "Paper-to-author relationships preserving complete author teams"

transform T_characterise_relations :
  intent: characterise
  manner: "define-relations (fractionally weighted co-authorship)"
  input: D_papers, D_authors, F_authorship
  output: F_coauthorship, F_text_representation
  actor: machine
  description: "Compute weighted co-authorship relations while preventing
    large author teams from dominating, and encode paper text for topic analysis"

artifact F_coauthorship : feature(D_authors)
  value structure: relational configuration
  value type: numeric
  representation form: "weighted temporal co-authorship network"
  description: "Author collaboration ties with fractional paper contributions
    and time references"

artifact F_text_representation : feature(D_papers)
  value structure: matrix
  value type: numeric
  representation form: "document-term representation"
  description: "Machine-processable representation of normalised titles,
    abstracts, and keywords"

transform T_define_snapshots :
  intent: define-unit
  manner: "time-partitioning and network aggregation"
  input: D_papers, D_authors, F_coauthorship, S_temporal_views
  output: D_network_snapshots
  actor: machine
  description: "Construct annual, rolling-window, and broad-era weighted
    co-authorship network snapshots"

artifact D_network_snapshots : entities
  internal structure: formation
  embedment: time
  features:
    - id: window_extent
      value structure: atomic
      value type: temporal
      description: "Time interval represented by the snapshot"
    - id: edge_weight
      value structure: relational configuration
      value type: numeric
      description: "Fractionally weighted co-authorship network"
  description: "Time-indexed co-authorship network snapshots at multiple scales"

assign:
  S_community_current := S_community_analysis

# ============================================================================
# OVERLAPPING COMMUNITY DISCOVERY AND REFINEMENT
# ============================================================================

loop L_community_refinement:
  purpose: "Find an interpretable and stable overlapping community structure
    while allowing resolution, weighting, and temporal choices to be revised"
  until: "The selected communities are sufficiently cohesive, stable across
    adjacent periods, topically interpretable, and credible to the analyst"
  body:

    transform T_fit_community_model :
      intent: build-model
      manner: "fit overlapping community-detection model"
      input: D_network_snapshots, F_coauthorship, S_community_current
      output: M_community_model
      actor: machine
      description: "Fit a candidate overlapping community model in each
        temporal network snapshot at the current resolution and settings"

    artifact M_community_model : model(D_network_snapshots, F_coauthorship)
      model type: "overlapping community-detection model"
      representation form: "membership-weight and network-partition parameters"
      description: "Computational representation of candidate overlapping
        author communities across temporal snapshots"

    transform T_define_communities :
      intent: define-unit
      manner: "derive overlapping community groups from membership weights"
      input: D_network_snapshots, D_authors, M_community_model
      output: D_communities, F_community_membership
      actor: machine
      description: "Define community instances for each time window and
        assign authors weighted memberships in one or more communities"

    artifact D_communities : entities
      internal structure: group/cluster
      embedment: time
      features:
        - id: active_window
          value structure: atomic
          value type: temporal
          description: "Time window in which the community instance is active"
      description: "Overlapping author communities observed in temporal snapshots"

    artifact F_community_membership : feature(D_authors)
      value structure: matrix
      value type: numeric
      representation form: "author × community-time membership weights"
      description: "Degree of each author's membership in each community
        over time"

    transform T_characterise_communities :
      intent: characterise
      manner: "compute core-periphery, leadership, and structural metrics"
      input: D_communities, D_authors, F_community_membership,
             F_coauthorship
      output: F_author_roles, F_community_metrics
      actor: machine
      description: "Identify candidate stable-core members, leaders,
        peripheral contributors, and boundary spanners using persistence,
        within-community collaboration, brokerage, productivity, and impact"

    artifact F_author_roles : feature(D_authors)
      value structure: vector
      value type: {categorical, numeric, temporal}
      description: "Time-resolved stable-core, leader, peripheral, and
        boundary-spanner roles with supporting scores"

    artifact F_community_metrics : feature(D_communities)
      value structure: vector
      value type: {numeric, temporal}
      description: "Community size, density, core persistence, overlap,
        repeated collaboration, publication volume, and impact indicators"

    transform T_contextualise_community_states :
      intent: contextualise
      manner: "projection-based dynamic network-state map"
      input: D_network_snapshots, F_community_metrics
      output: D_state_space, A_snapshot_states
      actor: machine
      description: "Project temporal network snapshots into a state space
        where proximity reflects similarity of community-network structure"

    artifact D_state_space : entities
      internal structure: elementary
      features:
        - id: dimensions
          value structure: atomic
          value type: numeric
          description: "Two dimensions of the computed state space"
      description: "Computed projection context for comparing network states"

    artifact A_snapshot_states : arrangement(D_network_snapshots)
      context: D_state_space
      principle: "similarity-preserving projection of network-state features"
      description: "Arrangement of temporal network snapshots by structural similarity"

    transform T_visualise_communities :
      intent: visualise
      manner: "coordinated community-structure views"
      input: D_communities, D_authors, F_community_membership,
             F_author_roles, F_community_metrics, A_snapshot_states
      output: V_community_structure
      actor: machine
      description: "Provide linked community views: temporal state map,
        community-size summaries, core-periphery profiles, membership
        histories, and NodeTrix-style matrix-plus-node-link inspection"

    artifact V_community_structure : visualisation(D_communities, D_authors,
        F_community_membership, F_author_roles, F_community_metrics,
        A_snapshot_states)
      layout: "coordinated state projection, temporal summaries, membership
        matrix, and selected-community matrix-plus-node-link view"
      form: "projected points, timelines, ribbons, matrices, and node-link diagrams"
      encoding: "community and author colour encode membership; matrix cells
        and links encode collaboration strength; size encodes membership or
        publication volume; time encodes snapshot order"
      description: "Interactive visual interface for inspecting candidate
        communities, core members, leaders, and cross-community ties"

    transform T_interpret_community_patterns :
      intent: abstract
      manner: "perception-and-interpretation of candidate communities"
      input: V_community_structure, F_author_roles, F_community_metrics,
             K_domain_knowledge
      output: P_candidate_community_structure
      actor: human
      description: "Interpret salient stable communities, possible
        subcommunities, unusual overlaps, and plausible leader/core roles"

    artifact P_candidate_community_structure : pattern(D_communities)
      representation form: "annotated community and role descriptions"
      description: "Analyst interpretations of cohesive communities,
        cores, leaders, boundary spanners, and suspicious fragmentations"

    transform T_assess_communities :
      intent: assess
      manner: "evaluate community quality, stability, and interpretability"
      input: D_communities, F_community_membership, F_community_metrics,
             V_community_structure, P_candidate_community_structure,
             S_community_current, K_domain_knowledge
      output: K_community_assessment, S_community_revised
      actor: hybrid
      description: "Evaluate cohesion, separation, parameter stability,
        persistence, and topical plausibility; determine whether community
        settings require further refinement"

    artifact K_community_assessment : knowledge(D_communities,
        F_community_metrics)
      representation form: "quality judgment"
      description: "Assessment of whether the community solution is
        fragmented, overly merged, unstable, or sufficiently interpretable"

    artifact S_community_revised : specification
      representation form: "revised configuration schema"
      description: "Updated community resolution, temporal-view selection,
        tie weighting, filtering, or model choice for another iteration"

    if K_community_assessment indicates further refinement needed:
      then:
        assign:
          S_community_current := S_community_revised
      else:
        exit loop L_community_refinement

end loop L_community_refinement

# ============================================================================
# TOPIC CHARACTERISATION AND HUMAN-STEERED TOPIC REFINEMENT
# ============================================================================

transform T_characterise_community_corpora :
  intent: characterise
  manner: "aggregate weighted document corpora"
  input: D_papers, D_communities, F_community_membership,
         F_text_representation
  output: F_community_corpora
  actor: machine
  description: "Construct weighted community-time document corpora.
    Contributions from authors belonging to several communities are
    fractionally allocated while preserving links to full papers."

artifact F_community_corpora : feature(D_communities)
  value structure: matrix
  value type: numeric
  representation form: "weighted community-time document-term matrix"
  description: "Textual evidence associated with each community instance"

assign:
  S_topic_current := S_topic_analysis

loop L_topic_refinement:
  purpose: "Develop interpretable topic profiles for communities and
    refine topic representations with analyst guidance"
  until: "Topics are distinct, meaningful to visualization researchers,
    and adequately explain communities' research foci and change over time"
  body:

    transform T_fit_topic_model :
      intent: build-model
      manner: "fit interpretable topic model"
      input: F_community_corpora, F_text_representation, S_topic_current
      output: M_topic_model
      actor: machine
      description: "Fit an initial topic model to weighted paper text from
        community-time corpora"

    artifact M_topic_model : model(F_community_corpora, F_text_representation)
      model type: "topic model"
      representation form: "nonnegative matrix factorisation or comparable
        topic-term and topic-document matrices"
      description: "Model of research topics and their prevalence in
        community-time paper corpora"

    transform T_define_topics :
      intent: define-unit
      manner: "derive topic groups and community topic mixtures"
      input: M_topic_model, F_community_corpora
      output: D_topics, F_topic_mixture, F_topic_terms
      actor: machine
      description: "Define candidate topics, representative papers, terms,
        and the mixture of topics associated with each community over time"

    artifact D_topics : entities
      internal structure: group/cluster
      embedment: set
      description: "Candidate research topics represented by associated
        terms and papers"

    artifact F_topic_mixture : feature(D_communities)
      value structure: vector
      value type: numeric
      description: "Topic prevalence profile for each community-time instance"

    artifact F_topic_terms : feature(D_topics)
      value structure: list
      value type: text
      description: "Ranked representative terms and papers for each topic"

    transform T_characterise_topic_change :
      intent: characterise
      manner: "compute topic evolution and cross-community similarity"
      input: D_topics, F_topic_mixture, F_topic_terms
      output: F_topic_trends, F_topic_similarity
      actor: machine
      description: "Calculate persistent, emerging, declining, and shared
        topics, plus topic diversity and topical similarity between communities"

    artifact F_topic_trends : feature(D_communities)
      value structure: vector
      value type: {numeric, temporal}
      description: "Temporal topic prevalence, diversity, emergence, and decline"

    artifact F_topic_similarity : feature(D_communities)
      value structure: matrix
      value type: numeric
      representation form: "pairwise community topical-similarity matrix"
      description: "Similarity of communities' topic profiles"

    transform T_visualise_topics :
      intent: visualise
      manner: "topic timeline and document inspection"
      input: D_communities, D_topics, F_topic_mixture, F_topic_terms,
             F_topic_trends
      output: V_topics
      actor: machine
      description: "Show topic timelines, community-topic heatmaps, topic
        terms, representative papers, and paper-level textual evidence"

    artifact V_topics : visualisation(D_communities, D_topics,
        F_topic_mixture, F_topic_terms, F_topic_trends)
      layout: "aligned community timelines, topic heatmap, and linked paper inspector"
      form: "stacked area charts, heatmaps, ranked term lists, and document cards"
      encoding: "time determines horizontal position; colour identifies
        topics; area or intensity encodes topic prevalence"
      description: "Interactive interface for interpreting community research foci"

    transform T_interpret_topics :
      intent: abstract
      manner: "name and refine research-focus patterns"
      input: V_topics, D_topics, F_topic_terms, K_domain_knowledge
      output: P_research_foci
      actor: human
      description: "Name meaningful topics, identify redundant or mixed
        topics, and relate topic patterns to known visualization research areas"

    artifact P_research_foci : pattern(D_topics, D_communities)
      representation form: "named topic descriptions and temporal focus patterns"
      description: "Interpreted research foci characterising communities
        and their evolution"

    transform T_assess_topics :
      intent: assess
      manner: "evaluate topic coherence and analytical usefulness"
      input: M_topic_model, V_topics, P_research_foci, S_topic_current
      output: K_topic_assessment, S_topic_revised
      actor: human
      description: "Judge topic coherence, distinctness, coverage, and
        explanatory value; decide whether terms, seeds, topic number, or
        topic structure should be revised"

    artifact K_topic_assessment : knowledge(D_topics, M_topic_model)
      representation form: "quality judgment"
      description: "Assessment of topic-model adequacy and need for refinement"

    artifact S_topic_revised : specification
      representation form: "revised topic-model settings and constraints"
      description: "Analyst revisions such as topic merging or splitting,
        generic-term removal, seeded concepts, preprocessing changes, or
        revised topic count"

    if K_topic_assessment indicates topic refinement needed:
      then:
        assign:
          S_topic_current := S_topic_revised
      else:
        exit loop L_topic_refinement

end loop L_topic_refinement

# ============================================================================
# COMMUNITY LINEAGES, OVERLAP, AND EVOLUTION EVENTS
# ============================================================================

transform T_define_lineages :
  intent: define-unit
  manner: "match communities across adjacent temporal windows"
  input: D_communities, F_community_membership, F_author_roles,
         F_topic_mixture, F_topic_similarity, S_lineage_matching
  output: D_community_lineages, F_lineage_membership, F_event_candidates
  actor: hybrid
  description: "Match community instances through time, weighting
    stable-core overlap more strongly than peripheral overlap, and derive
    candidate continuation, growth, merge, split, birth, dissolution,
    temporary-disappearance, and reorientation events"

artifact D_community_lineages : entities
  internal structure: sequence
  embedment: set
  features:
    - id: lifespan
      value structure: atomic
      value type: temporal
      description: "Temporal span of the community lineage"
  description: "Community lineages linking related community instances over time"

artifact F_lineage_membership : feature(D_communities)
  value structure: relational configuration
  value type: reference
  representation form: "predecessor-successor lineage graph"
  description: "Links between community instances and their inferred lineages"

artifact F_event_candidates : feature(D_community_lineages)
  value structure: list
  value type: categorical
  description: "Candidate evolution events with affected predecessor and
    successor communities"

transform T_characterise_overlap_and_evidence :
  intent: characterise
  manner: "compute overlap, mobility, and supporting evidence"
  input: D_authors, D_communities, D_community_lineages,
         F_community_membership, F_author_roles, F_topic_mixture,
         F_event_candidates, D_papers
  output: F_overlap, F_member_transitions, F_event_evidence
  actor: machine
  description: "Compute pairwise membership overlap, shared authors,
    multi-community participation, author transitions, shared papers,
    topical changes, venue shifts, and available citation evidence"

artifact F_overlap : feature(D_communities)
  value structure: matrix
  value type: numeric
  representation form: "time-resolved pairwise community-overlap matrix"
  description: "Strength and temporal change of shared membership between communities"

artifact F_member_transitions : feature(D_authors)
  value structure: relational configuration
  value type: reference
  representation form: "author-to-community transition histories"
  description: "Researchers' memberships, mobility, and simultaneous
    participation in several communities"

artifact F_event_evidence : feature(D_community_lineages)
  value structure: vector
  value type: {numeric, categorical, temporal, text}
  description: "Structural, membership, topical, paper, venue, and
    citation evidence supporting each candidate event"

artifact D_lineage_timeline : entities
  origin: given
  internal structure: elementary
  embedment: time
  features:
    - id: date
      value structure: atomic
      value type: temporal
      description: "Time position for lineage visualisation"
  description: "Temporal reference context for community-lineage displays"

transform T_contextualise_lineages :
  intent: contextualise
  manner: "time-based lineage arrangement"
  input: D_communities, D_lineage_timeline, F_lineage_membership
  output: A_lineage_timeline
  actor: machine
  description: "Arrange community instances by time and lineage relationship
    for visual tracing of continuations, merges, and splits"

artifact A_lineage_timeline : arrangement(D_communities)
  context: D_lineage_timeline
  principle: "temporal position with lineage-preserving band placement"
  description: "Time-based arrangement of community instances within lineages"

transform T_visualise_evolution :
  intent: visualise
  manner: "coordinated lineage, overlap, and event-investigation views"
  input: A_lineage_timeline, D_community_lineages, F_event_candidates,
         F_event_evidence, F_overlap, F_member_transitions, F_author_roles,
         F_topic_trends
  output: V_community_evolution
  actor: machine
  description: "Show a lineage diagram, membership ribbons, overlap matrix,
    selected-community NodeTrix view, topic timelines, and evidence-linked
    paper inspector"

artifact V_community_evolution : visualisation(A_lineage_timeline,
        D_community_lineages, F_event_candidates, F_event_evidence,
        F_overlap, F_member_transitions, F_author_roles, F_topic_trends)
  layout: "community lineage bands with coordinated membership, overlap,
    topic, matrix-network, and paper-detail views"
  form: "alluvial bands, ribbons, heatmaps, matrices, node-link diagrams,
    timelines, and document cards"
  encoding: "band width encodes active membership or publication volume;
    links encode lineage relations; colours encode community/topic identity;
    highlighting exposes shared authors, event evidence, and selected papers"
  description: "Interactive visual workspace for examining community
    persistence, overlap, mobility, and evolutionary events"

transform T_review_evolution_events :
  intent: abstract
  manner: "perception-and-interpretation of lineage events"
  input: V_community_evolution, F_event_candidates, F_event_evidence,
         P_research_foci, K_domain_knowledge
  output: P_reviewed_evolution_events
  actor: human
  description: "Review candidate events against underlying members,
    collaboration structure, representative papers, research foci, venues,
    and citations; distinguish genuine events from sparse-data artifacts"

artifact P_reviewed_evolution_events : pattern(D_community_lineages)
  representation form: "reviewed event annotations"
  description: "Analyst-reviewed continuations, merges, splits, births,
    dissolutions, temporary disappearances, and topical reorientations"

transform T_formulate_event_explanations :
  intent: generate-knowledge
  manner: "formulate evidence-backed explanations"
  input: P_reviewed_evolution_events, P_research_foci, F_event_evidence,
         K_domain_knowledge
  output: K_evolution_explanations
  actor: human
  description: "Formulate cautious explanations of when and why community
    changes occurred, explicitly distinguishing evidence-backed hypotheses
    from causal claims"

artifact K_evolution_explanations : knowledge(P_reviewed_evolution_events,
    P_research_foci, F_event_evidence)
  representation form: "annotated explanatory statements"
  description: "Evidence-backed interpretations of community evolution,
    including likely structural and topical reasons for major events"

# ============================================================================
# ROBUSTNESS AND UNCERTAINTY ANALYSIS
# ============================================================================

assign:
  S_robustness_current := S_robustness_analysis

loop L_robustness_analysis:
  purpose: "Test whether community, lineage, and topical conclusions persist
    under reasonable alternative analytical choices"
  until: "The analyst has examined sufficient plausible alternatives and
    can distinguish robust findings from parameter-sensitive findings"
  body:

    transform T_fit_alternative_models :
      intent: build-model
      manner: "fit alternative community and topic models"
      input: D_network_snapshots, F_coauthorship, F_community_corpora,
             S_robustness_current
      output: M_alternative_community_model, M_alternative_topic_model
      actor: machine
      description: "Re-run community and topic modelling under selected
        alternative temporal windows, edge weights, filters, resolutions,
        preprocessing choices, and topic-model settings"

    artifact M_alternative_community_model : model(D_network_snapshots,
        F_coauthorship)
      model type: "alternative overlapping community-detection model"
      description: "One sensitivity-analysis community model configuration"

    artifact M_alternative_topic_model : model(F_community_corpora)
      model type: "alternative topic model"
      description: "One sensitivity-analysis topic model configuration"

    transform T_define_alternative_results :
      intent: define-unit
      manner: "derive alternative communities, topics, and lineages"
      input: D_network_snapshots, D_authors, M_alternative_community_model,
             M_alternative_topic_model, S_lineage_matching
      output: D_alternative_communities, D_alternative_lineages,
              F_alternative_membership, F_alternative_topic_mixture
      actor: machine
      description: "Derive comparable alternative community memberships,
        topic profiles, and lineage structures for sensitivity comparison"

    artifact D_alternative_communities : entities
      internal structure: group/cluster
      embedment: time
      description: "Community instances obtained under an alternative configuration"

    artifact D_alternative_lineages : entities
      internal structure: sequence
      embedment: set
      description: "Lineages obtained under an alternative configuration"

    artifact F_alternative_membership : feature(D_authors)
      value structure: matrix
      value type: numeric
      description: "Alternative author-community membership weights"

    artifact F_alternative_topic_mixture : feature(D_alternative_communities)
      value structure: vector
      value type: numeric
      description: "Alternative community topic-prevalence profiles"

    transform T_characterise_robustness :
      intent: characterise
      manner: "compare alternative analytical results"
      input: D_communities, D_community_lineages, F_community_membership,
             F_topic_mixture, D_alternative_communities,
             D_alternative_lineages, F_alternative_membership,
             F_alternative_topic_mixture
      output: F_result_robustness
      actor: machine
      description: "Measure agreement and sensitivity of community cores,
        memberships, leaders, lineages, events, and research foci across
        alternative analysis configurations"

    artifact F_result_robustness : feature(D_community_lineages)
      value structure: vector
      value type: {numeric, categorical}
      description: "Stability scores and robust/tentative/uncertain labels
        for communities, topics, and lineage events"

    transform T_visualise_robustness :
      intent: visualise
      manner: "comparative sensitivity display"
      input: F_result_robustness, D_communities, D_alternative_communities,
             D_community_lineages, D_alternative_lineages
      output: V_robustness
      actor: machine
      description: "Compare primary and alternative solutions, highlighting
        stable communities and events as well as parameter-sensitive results"

    artifact V_robustness : visualisation(F_result_robustness, D_communities,
        D_alternative_communities, D_community_lineages,
        D_alternative_lineages)
      layout: "aligned solution comparison and stability summaries"
      form: "matched community bands, agreement matrices, and ranked stability plots"
      encoding: "alignment links comparable results; colour and opacity
        encode robustness category and stability score"
      description: "Visual evidence for judging robustness and uncertainty"

    transform T_assess_robustness :
      intent: assess
      manner: "judge sensitivity-analysis sufficiency"
      input: V_robustness, F_result_robustness, S_robustness_current,
             K_domain_knowledge
      output: K_robustness_assessment, S_robustness_revised
      actor: human
      description: "Decide whether the tested alternatives sufficiently
        cover plausible analytical choices or whether additional settings
        require examination"

    artifact K_robustness_assessment : knowledge(F_result_robustness)
      representation form: "quality judgment"
      description: "Assessment of which findings are robust, tentative,
        uncertain, and whether further sensitivity analysis is needed"

    artifact S_robustness_revised : specification
      representation form: "revised parameter settings"
      description: "Additional alternative settings selected for robustness testing"

    if K_robustness_assessment indicates additional alternatives needed:
      then:
        assign:
          S_robustness_current := S_robustness_revised
      else:
        exit loop L_robustness_analysis

end loop L_robustness_analysis

# ============================================================================
# SYNTHESIS AND REPORTING
# ============================================================================

transform T_synthesise_findings :
  intent: generate-knowledge
  manner: "formulate analytical report"
  input: D_community_lineages, F_author_roles, F_community_metrics,
         P_research_foci, P_reviewed_evolution_events,
         K_evolution_explanations, K_robustness_assessment,
         F_result_robustness
  output: K_community_findings
  actor: human
  description: "Produce a documented account of the number of active
    communities and long-term lineages, their cores and leaders, research
    foci, overlap and mobility, evolutionary events, supporting evidence,
    and uncertainty"

artifact K_community_findings : knowledge(D_community_lineages,
    P_research_foci, P_reviewed_evolution_events, F_result_robustness)
  representation form: "annotated analytical report and provenance record"
  description: "Final evidence-linked findings about the intellectual
    structure and evolution of the IEEE VIS research community"

Structural elements made explicit

The ATWL representation makes several structures explicit that were implicit in the natural-language workflow:

• Specifications: temporal-window choices, community-detection settings, topic-model settings, lineage-matching rules, and robustness settings are represented as explicit specification artifacts.
• Human–machine roles: automated network/topic computation is marked as machine-driven; interpretation, quality judgments, and event explanations are marked as human or hybrid.
• Refinement loops: the original recommendation explicitly called for iterative community selection, human-steered topic refinement, and sensitivity testing. ATWL formalises these as three loops with assessment artifacts, revised specifications, and stopping conditions.
• Intermediate artifacts: co-authorship networks, weighted memberships, core/leader roles, community corpora, topic mixtures, lineages, overlaps, event evidence, and robustness scores are made explicit so that visual inspection and later conclusions have traceable inputs.
• Event review: the recommendation required analyst review of merges, splits, and dissolutions. ATWL distinguishes machine-derived event candidates from human-reviewed event patterns and from final explanatory knowledge.

No new substantive analytical phase was added. The separate robustness loop is a formal rendering of the recommendation to repeat detection and tracking under reasonable alternative parameter settings and flag parameter-sensitive conclusions.

Recommended modifications

Thinking in ATWL terms suggests several improvements that would strengthen the workflow itself.

1. Externalise quality criteria as explicit specifications
   The workflow names criteria such as cohesion, stability, interpretability, and topical plausibility, but does not fully define them in advance. Add explicit criteria for: - minimum community size and duration; - stable-core membership and persistence thresholds; - leader-ranking measures and their relative importance; - acceptable membership-overlap levels; - topic coherence and distinctness; - thresholds for continuation, merge, split, birth, and dissolution.

These criteria should remain adjustable, but recording them makes results more reproducible.

2. Create a joint community–topic refinement stage
   Community quality is partly judged by topical interpretability, yet detailed topic modelling occurs only after the community-detection loop. This creates a dependency gap.

Add either: - an initial lightweight topical profile before accepting a community solution; or - a joint loop in which community detection and topic characterisation are alternated.

This would allow the analyst to reject a structurally dense cluster that has no coherent intellectual focus, or detect a structurally diffuse but topically meaningful interdisciplinary community.

3. Add a lineage-validation and refinement loop
   The workflow generates candidate lineage events and asks the analyst to review them, but does not explicitly allow review to revise lineage-matching rules.

Add a loop in which the analyst can adjust: - core-overlap weighting; - topical-similarity weighting; - event thresholds; - rules for temporary disappearance; - treatment of sparse early periods.

This is especially important because merge and split labels can be highly sensitive to matching thresholds and window width.

4. Specify how “stable core” and “leader” are operationalised
   The workflow appropriately avoids equating leadership with citation count alone, but its composite role definitions should be made more explicit.

Add a role-definition specification covering: - persistence horizon, such as membership across at least (k) adjacent windows; - minimum membership weight; - repeated-collaboration requirements; - within-community versus cross-community centrality; - whether productivity and citations are supplementary indicators or part of a composite score.

This would make distinctions between core members, leaders, brokers, and prolific short-term contributors more defensible.

5. Formalise uncertainty and resampling procedures
   The workflow recommends testing alternative parameters and mentions bootstrap resampling, but the uncertainty procedure is not fully specified.

Add a robustness specification that defines: - resampling unit: papers, collaboration edges, authors, or time windows; - number of resamples; - agreement measures for community memberships and cores; - lineage-event confidence measures; - criteria for labelling results as robust, tentative, or uncertain.

6. Separate data-cleaning judgments from cleaning policy
   The audit produces analyst judgments, but the resulting cleaning and inclusion rules should be captured as an explicit, versioned policy. For example: - duplicate-paper resolution rule; - author-team-size treatment; - missing-abstract handling; - inclusion of isolated authors; - citation-data treatment; - handling of venue coverage changes.

This would make the transition from raw records to the analytical corpus easier to reproduce.

7. Clarify citation semantics before analysis
   The workflow correctly distinguishes citation links from citation counts, but this is central enough to require an explicit initial decision. Citation counts can indicate impact; citation links can support intellectual-flow analysis. They should not be substituted for each other.

Add a citation-availability assessment and a downstream analysis specification stating whether citation analysis is: - unavailable; - limited to paper impact; - based on within-corpus citation links; or - based on a broader external citation graph.

8. Specify attribution for multi-community papers
   Fractional allocation of papers is sensible, but the allocation rule can substantially affect topic profiles. Add a specification for whether a paper’s contribution is allocated by: - the contributing author’s community-membership weight; - equal fractions across relevant communities; - full membership in every associated community; - a primary-community rule for selected reporting views.

The analyst should be able to compare fractional and full-membership interpretations for interdisciplinary work.

9. Add an explicit provenance record throughout, not only at reporting
   The workflow recommends maintaining analysis history, but this should be an active artifact updated after major assessments and refinements.

The record should capture: - specification versions; - parameter changes; - model outputs compared; - topic edits; - manual event annotations; - reasons for accepting or rejecting alternatives; - links to supporting papers and visual evidence.

10. Use an evidence hierarchy for explanations of “why” events occurred
    Bibliographic and network data can show that a merge, split, or reorientation coincided with changes in people, topics, venues, and citations. They generally cannot establish causality by themselves.

Add explicit interpretive categories such as: - descriptive structural observation; - evidence-backed hypothesis; - externally corroborated explanation.

If the analyst needs stronger causal explanations, augment the corpus with external evidence, such as conference-history records, institutional affiliations, funding initiatives, interviews, or major technical developments.

These modifications do not change the central workflow goal. They primarily make its iterative decisions, uncertainty treatment, and evidential basis more explicit and auditable.

Refining community topics

I’m thinking about possibly revising the community config and quality criteria together. It seems important to have an initial thematic screening in our community loop. We could build a lightweight topic model with each iteration, defining lightweight topics and assessing them. After that, we’d engage in a more detailed human-steered topic loop. The full topic model would depend on the accepted community. I’ll also explore using text corpora from candidate communities to compare settings.

Assessing community coherence

I'm thinking about how to characterize topic coherence and visualize communities related to that. I want to assess the needs for quality and consider stable roles with appropriate definitions. It seems essential to define roles explicitly, but I wonder if I should revise them based on audit findings? While interpretation risks post-hoc adjustments, it's crucial to ensure definitions capture the intended concepts. I should also think about the lineage validation loop and how the community results might be altered.

Comparing outputs and roles

I need to compare different aspects, like the attribution that includes primary reporting and sensitivity rules. In the topic model, I'm considering how to use various corpora and ensure a robust analysis loop with alternative allocations. It could be interesting to explore outputs from alternative communities and lineages while defining core roles for comparison. I want to characterize and define events or roles more clearly, ensuring that provenance is updated throughout the process with accurate data policies in mind.

Configuring knowledge provenance

I need to clarify that provenance doesn't directly configure transforms. Knowledge can act as a knowledge artifact, but it must be generated. I think using a "K_provenance_seed" origin defined as "empty protocol" would help. Each phase should output a "K_provenance..." and assign it as the current record. This process might generate knowledge with a primary intent while assessing transforms. It's good to include an audit policy, but not every loop update has to be included as output. I should focus on configuring K provenance records and annotations.

workflow ieee-vis-community-evolution-revised
  template: characterise -> visualise -> assess -> generate-knowledge ->
            define-unit -> characterise -> define-unit ->
            loop(build-model -> define-unit -> characterise ->
                 build-model -> define-unit -> characterise ->
                 contextualise -> visualise -> abstract -> assess) ->
            characterise ->
            loop(build-model -> define-unit -> characterise ->
                 visualise -> abstract -> assess) ->
            loop(define-unit -> characterise -> contextualise ->
                 visualise -> abstract -> assess) ->
            loop(define-unit -> build-model -> define-unit ->
                 characterise -> build-model -> define-unit ->
                 define-unit -> characterise -> visualise -> assess) ->
            generate-knowledge
  description: "Identify and interpret evolving, overlapping collaboration
    communities in IEEE VIS publications. The workflow characterises
    community cores, leaders, topical foci, membership overlap, and
    lineage events while documenting methodological decisions and
    uncertainty."

# ============================================================================
# INPUT DATA, DOMAIN KNOWLEDGE, AND INITIAL CONTROL SPECIFICATIONS
# ============================================================================

artifact D_papers_raw : entities
  origin: given
  internal structure: elementary
  embedment: {set, time}
  features:
    - id: paper_id
      value structure: atomic
      value type: reference
      description: "Paper identifier"
    - id: publication_year
      value structure: atomic
      value type: temporal
      description: "Publication year, 1990--2024"
    - id: venue
      value structure: atomic
      value type: categorical
      description: "VIS, InfoVis, VAST, or SciVis"
    - id: authors
      value structure: list
      value type: reference
      description: "DBLP-disambiguated author identifiers and raw names"
    - id: bibliographic_text
      value structure: vector
      value type: text
      description: "Title, abstract, and keywords"
    - id: citation_information
      value structure: relational configuration
      value type: {numeric, reference}
      description: "Citation links where available, otherwise citation counts"
  description: "Raw IEEE VIS conference-paper records"

artifact K_domain_knowledge : knowledge(D_papers_raw)
  origin: given
  representation form: "annotations, taxonomies, and interpretive criteria"
  description: "Visualization-domain knowledge used to interpret research
    foci, venue history, plausible collaboration structures, and events"

artifact K_provenance_seed : knowledge(D_papers_raw)
  origin: given
  representation form: "empty provenance protocol"
  description: "Initial protocol for recording specifications, revisions,
    visual evidence, analyst annotations, and reasons for decisions"

artifact S_temporal_views : specification
  origin: given
  representation form: "configuration schema"
  description: "Candidate annual, rolling three- or five-year, and
    broad-era temporal views"

artifact S_community_analysis : specification
  origin: given
  representation form: "configuration schema"
  description: "Initial co-authorship weighting, overlapping-community
    method, resolution range, weak-tie treatment, and candidate
    temporal-window settings"

artifact S_quality_criteria : specification
  origin: given
  representation form: "quality criteria and thresholds"
  description: "Initial criteria for minimum community size and duration,
    cohesion, separation, persistence, topical coherence, and
    interpretability"

artifact S_role_definition : specification
  origin: given
  representation form: "role-definition rules"
  description: "Rules defining stable-core members, leaders, peripheral
    contributors, and boundary spanners; includes persistence horizon,
    membership threshold, repeated-collaboration requirement, and the
    relative use of centrality, productivity, brokerage, and impact"

artifact S_topic_analysis : specification
  origin: given
  representation form: "configuration schema"
  description: "Text preprocessing, generic-term treatment, topic-model
    family, topic-count range, coherence criteria, and permitted edits"

artifact S_multi_community_attribution : specification
  origin: given
  representation form: "allocation rules"
  description: "Rules for allocating papers associated with authors who
    belong to multiple communities: weighted fractional allocation,
    full-membership evidence view, and primary-community reporting view"

artifact S_lineage_matching : specification
  origin: given
  representation form: "matching rules and thresholds"
  description: "Rules for continuation, growth, merge, split, birth,
    dissolution, temporary disappearance, and reorientation; includes
    relative weighting of stable-core overlap, total membership overlap,
    topical similarity, and collaboration-structure similarity"

artifact S_robustness_analysis : specification
  origin: given
  representation form: "resampling and sensitivity settings"
  description: "Alternative time windows, tie weights, filters, resolutions,
    attribution rules, topic settings, resampling unit, number of
    resamples, agreement measures, and robust/tentative/uncertain labels"

artifact S_explanation_standard : specification
  origin: given
  representation form: "evidence hierarchy"
  description: "Rules distinguishing descriptive structural observations,
    evidence-backed hypotheses, and externally corroborated explanations;
    prohibits causal claims from bibliographic evidence alone"

assign:
  K_provenance_current := K_provenance_seed

# ============================================================================
# DATA AUDIT, CITATION-SEMANTICS DECISION, AND CLEANING POLICY
# ============================================================================

transform T_audit_records :
  intent: characterise
  manner: "compute data quality and citation-availability statistics"
  input: D_papers_raw
  output: F_record_quality, F_citation_availability
  actor: machine
  description: "Identify duplicate papers, missing author or text fields,
    unusually large author teams, yearly venue coverage, and whether
    citations are links, counts, both, or unavailable"

artifact F_record_quality : feature(D_papers_raw)
  value structure: vector
  value type: {numeric, categorical}
  description: "Completeness, duplication, author-team-size, and venue-
    coverage indicators"

artifact F_citation_availability : feature(D_papers_raw)
  value structure: vector
  value type: categorical
  description: "Availability and semantics of citation links and counts"

transform T_visualise_audit :
  intent: visualise
  manner: "temporal coverage and quality overview"
  input: D_papers_raw, F_record_quality, F_citation_availability
  output: V_data_audit
  actor: machine
  description: "Show yearly publication counts by venue, missingness,
    duplicates, team sizes, and citation-data availability"

artifact V_data_audit : visualisation(D_papers_raw, F_record_quality,
    F_citation_availability)
  layout: "aligned venue timelines and linked quality summaries"
  form: "stacked bars, distributions, and record inspector"
  encoding: "time and venue determine publication-count displays; colour
    and highlighting identify quality and citation-data conditions"
  description: "Audit interface for corpus coverage and record quality"

transform T_assess_audit :
  intent: assess
  manner: "judge corpus adequacy and citation semantics"
  input: V_data_audit, F_record_quality, F_citation_availability,
         K_domain_knowledge, S_quality_criteria
  output: K_audit_assessment
  actor: human
  description: "Assess data quality, treatment of duplicates and missing
    fields, effects of venue coverage, and the appropriate analytical use
    of available citation information"

artifact K_audit_assessment : knowledge(D_papers_raw, F_record_quality,
    F_citation_availability)
  representation form: "quality judgment and data-treatment decisions"
  description: "Analyst judgment of corpus adequacy and citation semantics"

transform T_formulate_data_policy :
  intent: generate-knowledge
  manner: "formulate methodological policy and provenance entry"
  input: K_audit_assessment, K_provenance_current
  output: S_cleaning_policy, S_citation_scope, K_provenance_after_audit
  actor: human
  description: "Externalise reproducible cleaning and inclusion rules and
    decide whether citations are excluded, used only as impact indicators,
    or analysed as citation links"

artifact S_cleaning_policy : specification
  representation form: "data-cleaning and inclusion rules"
  description: "Versioned policy for duplicate resolution, missing-text
    handling, author-team-size treatment, isolated-author inclusion, and
    venue-coverage qualifications"

artifact S_citation_scope : specification
  representation form: "citation-analysis method choice"
  description: "Decision specifying whether citation information is used as
    impact only, as within-corpus links, as external links, or not used"

artifact K_provenance_after_audit : knowledge(K_audit_assessment,
    S_cleaning_policy, S_citation_scope)
  representation form: "provenance record"
  description: "Recorded audit evidence and adopted data and citation policy"

assign:
  K_provenance_current := K_provenance_after_audit

transform T_prepare_records :
  intent: define-unit
  manner: "filter, deduplicate, and normalise"
  input: D_papers_raw, S_cleaning_policy
  output: D_papers
  actor: machine
  description: "Create an analysis-ready corpus while retaining DBLP
    identifiers, raw display names, clean bibliographic text, and policy-
    compliant paper records"

artifact D_papers : entities
  internal structure: elementary
  embedment: {set, time}
  features:
    - id: publication_year
      value structure: atomic
      value type: temporal
      description: "Validated publication year"
    - id: venue
      value structure: atomic
      value type: categorical
      description: "Publication venue"
    - id: author_ids
      value structure: list
      value type: reference
      description: "Validated DBLP author identifiers"
    - id: clean_text
      value structure: vector
      value type: text
      description: "Normalised title, abstract, and keywords"
  description: "Prepared paper corpus"

# ============================================================================
# AUTHOR NETWORK AND TEMPORAL SNAPSHOTS
# ============================================================================

transform T_define_authors :
  intent: define-unit
  manner: "group-by-entity"
  input: D_papers
  output: D_authors, F_authorship
  actor: machine
  description: "Define unique researchers and preserve paper--author
    affiliations"

artifact D_authors : entities
  internal structure: group/cluster
  embedment: set
  features:
    - id: author_id
      value structure: atomic
      value type: reference
      description: "DBLP author identifier"
    - id: display_names
      value structure: list
      value type: text
      description: "Raw names retained for display and checking"
  description: "Researchers represented through associated papers"

artifact F_authorship : feature(D_papers)
  value structure: relational configuration
  value type: reference
  representation form: "paper--author bipartite network"
  description: "Complete authorship relationships"

transform T_characterise_relations :
  intent: characterise
  manner: "define fractional co-authorship relations and encode text"
  input: D_papers, D_authors, F_authorship
  output: F_coauthorship, F_text_representation
  actor: machine
  description: "Compute fractionally weighted temporal co-authorship ties
    and a machine-processable document representation"

artifact F_coauthorship : feature(D_authors)
  value structure: relational configuration
  value type: numeric
  representation form: "weighted temporal co-authorship network"
  description: "Co-authorship ties whose contribution is normalised within
    each paper's author team"

artifact F_text_representation : feature(D_papers)
  value structure: matrix
  value type: numeric
  representation form: "document-term matrix"
  description: "Text representation of normalised bibliographic content"

transform T_define_snapshots :
  intent: define-unit
  manner: "time-partitioning and network aggregation"
  input: D_papers, D_authors, F_coauthorship, S_temporal_views
  output: D_network_snapshots
  actor: machine
  description: "Construct annual, rolling-window, and broad-era weighted
    co-authorship network snapshots"

artifact D_network_snapshots : entities
  internal structure: formation
  embedment: time
  features:
    - id: window_extent
      value structure: atomic
      value type: temporal
      description: "Snapshot time interval"
    - id: edge_weight
      value structure: relational configuration
      value type: numeric
      description: "Weighted co-authorship ties"
  description: "Temporal collaboration-network snapshots"

assign:
  S_community_current := S_community_analysis
  S_quality_current := S_quality_criteria
  S_role_current := S_role_definition

# ============================================================================
# JOINT COMMUNITY-STRUCTURE AND LIGHTWEIGHT TOPICAL-SCREENING LOOP
# ============================================================================

loop L_community_refinement:
  purpose: "Find overlapping communities that are structurally coherent,
    temporally persistent, topically intelligible, and useful to the
    analyst"
  until: "The community structure satisfies the current quality criteria,
    including structural and lightweight topical-screening evidence"
  body:

    transform T_fit_community_model :
      intent: build-model
      manner: "fit overlapping community-detection model"
      input: D_network_snapshots, F_coauthorship, S_community_current
      output: M_community_model
      actor: machine
      description: "Fit candidate overlapping community memberships for
        each temporal network snapshot"

    artifact M_community_model : model(D_network_snapshots, F_coauthorship)
      model type: "overlapping community-detection model"
      representation form: "membership-weight and resolution parameters"
      description: "Candidate dynamic overlapping community structure"

    transform T_define_communities :
      intent: define-unit
      manner: "derive weighted overlapping groups"
      input: D_network_snapshots, D_authors, M_community_model
      output: D_communities, F_community_membership
      actor: machine
      description: "Define community instances and author membership weights
        for each temporal window"

    artifact D_communities : entities
      internal structure: group/cluster
      embedment: time
      description: "Candidate overlapping author communities"

    artifact F_community_membership : feature(D_authors)
      value structure: matrix
      value type: numeric
      representation form: "author x community-time membership weights"
      description: "Strength of every author's membership in communities
        across time"

    transform T_characterise_community_structure :
      intent: characterise
      manner: "compute role, persistence, and network metrics"
      input: D_communities, D_authors, F_community_membership,
             F_coauthorship, S_role_current
      output: F_author_roles, F_community_metrics
      actor: machine
      description: "Compute candidate stable cores, leaders, peripheral
        contributors, and boundary spanners according to explicit role rules"

    artifact F_author_roles : feature(D_authors)
      value structure: vector
      value type: {categorical, numeric, temporal}
      description: "Time-resolved role labels and supporting persistence,
        within-community, brokerage, productivity, and impact measures"

artifact F_community_metrics : feature(D_communities)
  value structure: vector
  value type: {numeric, temporal}
  description: "Size, density, separation, persistence, repeated-
    collaboration, overlap, productivity, and impact indicators"

    transform T_characterise_candidate_corpora :
      intent: characterise
      manner: "aggregate provisional community document corpora"
      input: D_papers, D_communities, F_community_membership,
             F_text_representation, S_multi_community_attribution
      output: F_candidate_community_corpora
      actor: machine
      description: "Create provisional weighted document corpora for
        topical screening of candidate communities"

    artifact F_candidate_community_corpora : feature(D_communities)
      value structure: matrix
      value type: numeric
      representation form: "weighted community-time document-term matrix"
      description: "Provisional textual evidence for each candidate community"

    transform T_fit_screening_topic_model :
      intent: build-model
      manner: "fit lightweight topical-screening model"
      input: F_candidate_community_corpora, S_topic_analysis
      output: M_screening_topic_model
      actor: machine
      description: "Estimate a lightweight topic representation used only
        to assess whether candidate communities have intelligible foci"

    artifact M_screening_topic_model : model(F_candidate_community_corpora)
      model type: "topic model"
      representation form: "compact topic-term and topic-community matrices"
      description: "Provisional topic model for community-solution assessment"

    transform T_define_screening_topics :
      intent: define-unit
      manner: "derive provisional topic groups"
      input: M_screening_topic_model, F_candidate_community_corpora
      output: D_screening_topics, F_screening_topic_profile
      actor: machine
      description: "Define provisional topics and their prevalence in
        candidate communities"

    artifact D_screening_topics : entities
      internal structure: group/cluster
      embedment: set
      description: "Provisional research-focus groups for screening"

    artifact F_screening_topic_profile : feature(D_communities)
      value structure: vector
      value type: numeric
      description: "Provisional topic prevalence and topical concentration
        per community-time instance"

    transform T_contextualise_community_states :
      intent: contextualise
      manner: "projection-based dynamic network-state map"
      input: D_network_snapshots, F_community_metrics
      output: D_state_space, A_snapshot_states
      actor: machine
      description: "Project snapshots so proximity represents similarity of
        their community-network structure"

    artifact D_state_space : entities
      internal structure: elementary
      features:
        - id: dimensions
          value structure: atomic
          value type: numeric
          description: "Two projection dimensions"
      description: "Computed network-state projection context"

    artifact A_snapshot_states : arrangement(D_network_snapshots)
      context: D_state_space
      principle: "similarity-preserving projection of network and community metrics"
      description: "Arrangement of temporal snapshots by structural similarity"

    transform T_visualise_community_candidates :
      intent: visualise
      manner: "coordinated community, role, and topical-screening views"
      input: D_communities, D_authors, F_community_membership,
             F_author_roles, F_community_metrics, A_snapshot_states,
             F_screening_topic_profile
      output: V_community_candidates
      actor: machine
      description: "Provide linked state-map, membership-history, core-
        periphery, NodeTrix-style, and provisional topical-profile views"

    artifact V_community_candidates : visualisation(D_communities, D_authors,
        F_community_membership, F_author_roles, F_community_metrics,
        A_snapshot_states, F_screening_topic_profile)
      layout: "state projection, temporal membership matrix, core-periphery
        summaries, selected-community matrices, and topical summaries"
      form: "projected points, ribbons, matrices, node-link diagrams,
        distributions, and topic bars"
      encoding: "colour denotes communities or topics; size and edge weight
        encode membership and collaboration strength; time orders snapshots"
      description: "Interactive workspace for inspecting candidate
        communities and their provisional research foci"

    transform T_interpret_community_patterns :
      intent: abstract
      manner: "perception-and-interpretation"
      input: V_community_candidates, F_author_roles, F_community_metrics,
             F_screening_topic_profile, K_domain_knowledge
      output: P_candidate_community_structure
      actor: human
      description: "Interpret salient communities, possible fragmentation,
        implausible merges, core and leader roles, overlap, and preliminary
        topical intelligibility"

    artifact P_candidate_community_structure : pattern(D_communities)
      representation form: "annotated candidate community descriptions"
      description: "Analyst interpretations of community structure and
        provisional intellectual coherence"

    transform T_assess_communities :
      intent: assess
      manner: "evaluate structural, temporal, role, and topical adequacy"
      input: D_communities, F_community_membership, F_author_roles,
             F_community_metrics, F_screening_topic_profile,
             V_community_candidates, P_candidate_community_structure,
             S_community_current, S_quality_current, S_role_current,
             K_domain_knowledge, K_provenance_current
      output: K_community_assessment, S_community_revised,
              S_quality_revised, S_role_revised,
              K_provenance_after_community_assessment
      actor: hybrid
      description: "Evaluate candidate communities against explicit
        thresholds and interpretive criteria; determine whether model
        settings, quality criteria, or role rules need justified revision"

    artifact K_community_assessment : knowledge(D_communities,
        F_community_metrics, F_screening_topic_profile)
      representation form: "quality judgment"
      description: "Assessment of cohesion, persistence, overlap,
        interpretability, topical intelligibility, and role validity"

    artifact S_community_revised : specification
      representation form: "revised community-analysis settings"
      description: "Justified revisions to method choice, resolution,
        temporal-window choice, weighting, or weak-tie treatment"

    artifact S_quality_revised : specification
      representation form: "revised quality criteria and thresholds"
      description: "Documented revisions to quality criteria when initial
        thresholds prove inappropriate for the corpus or historical period"

    artifact S_role_revised : specification
      representation form: "revised role-definition rules"
      description: "Documented revisions to stable-core, leader, peripheral,
        or broker definitions when evidence shows they are inadequate"

    artifact K_provenance_after_community_assessment : knowledge(
        K_community_assessment, S_community_revised, S_quality_revised,
        S_role_revised)
      representation form: "provenance record"
      description: "Recorded community-model settings, evidence, assessment,
        and rationale for accepted or revised choices"

    if K_community_assessment indicates refinement needed:
      then:
        assign:
          S_community_current := S_community_revised
          S_quality_current := S_quality_revised
          S_role_current := S_role_revised
          K_provenance_current := K_provenance_after_community_assessment
      else:
        assign:
          K_provenance_current := K_provenance_after_community_assessment
        exit loop L_community_refinement

end loop L_community_refinement

# ============================================================================
# DETAILED HUMAN-STEERED TOPIC CHARACTERISATION
# ============================================================================

transform T_characterise_community_corpora :
  intent: characterise
  manner: "aggregate weighted and full-membership document corpora"
  input: D_papers, D_communities, F_community_membership,
         F_text_representation, S_multi_community_attribution
  output: F_community_corpora, F_full_membership_papers
  actor: machine
  description: "Build topic-model corpora using the selected fractional
    allocation rule while retaining full paper--community associations for
    inspection of interdisciplinary work"

artifact F_community_corpora : feature(D_communities)
  value structure: matrix
  value type: numeric
  representation form: "weighted community-time document-term matrix"
  description: "Text corpus allocated to community-time instances"

artifact F_full_membership_papers : feature(D_communities)
  value structure: relational configuration
  value type: reference
  representation form: "full paper-community association graph"
  description: "Unfractionated evidence linking papers to every associated
    community"

assign:
  S_topic_current := S_topic_analysis
  S_attribution_current := S_multi_community_attribution

loop L_topic_refinement:
  purpose: "Develop meaningful and traceable research-focus profiles for
    accepted communities"
  until: "Topics satisfy coherence and distinctness criteria, are meaningful
    to the analyst, and adequately describe community focus and change"
  body:

    transform T_fit_topic_model :
      intent: build-model
      manner: "fit interpretable topic model"
      input: F_community_corpora, F_text_representation, S_topic_current
      output: M_topic_model
      actor: machine
      description: "Fit an interpretable topic model to community-associated
        paper text"

    artifact M_topic_model : model(F_community_corpora, F_text_representation)
      model type: "topic model"
      representation form: "topic-term and topic-document matrices"
      description: "Model of research foci in the accepted communities"

    transform T_define_topics :
      intent: define-unit
      manner: "derive topic groups and topic mixtures"
      input: M_topic_model, F_community_corpora
      output: D_topics, F_topic_mixture, F_topic_terms
      actor: machine
      description: "Define topics, representative terms and papers, and
        community-time topic mixtures"

    artifact D_topics : entities
      internal structure: group/cluster
      embedment: set
      description: "Candidate research topics"

    artifact F_topic_mixture : feature(D_communities)
      value structure: vector
      value type: numeric
      description: "Topic prevalence profile for each community over time"

    artifact F_topic_terms : feature(D_topics)
      value structure: list
      value type: text
      description: "Ranked terms and representative papers for each topic"

    transform T_characterise_topic_change :
      intent: characterise
      manner: "compute topic trends and community similarity"
      input: D_topics, F_topic_mixture, F_topic_terms
      output: F_topic_trends, F_topic_similarity
      actor: machine
      description: "Compute persistent, emerging, declining, and shared
        topics, topical diversity, and community topical similarity"

    artifact F_topic_trends : feature(D_communities)
      value structure: vector
      value type: {numeric, temporal}
      description: "Topic prevalence, diversity, emergence, and decline"

    artifact F_topic_similarity : feature(D_communities)
      value structure: matrix
      value type: numeric
      representation form: "pairwise community topical-similarity matrix"
      description: "Topical similarity among community-time instances"

    transform T_visualise_topics :
      intent: visualise
      manner: "topic timeline and paper-evidence inspection"
      input: D_communities, D_topics, F_topic_mixture, F_topic_terms,
             F_topic_trends, F_full_membership_papers
      output: V_topics
      actor: machine
      description: "Show topic timelines, community-topic heatmaps, terms,
        representative papers, and full-membership evidence"

    artifact V_topics : visualisation(D_communities, D_topics,
        F_topic_mixture, F_topic_terms, F_topic_trends,
        F_full_membership_papers)
      layout: "aligned community timelines, topic heatmap, and paper inspector"
      form: "stacked areas, heatmaps, ranked lists, and document cards"
      encoding: "time determines horizontal position; topic colour and
        area or intensity encode topic prevalence"
      description: "Interactive interface for interpreting research foci"

    transform T_interpret_topics :
      intent: abstract
      manner: "name and refine research-focus patterns"
      input: V_topics, D_topics, F_topic_terms, K_domain_knowledge
      output: P_research_foci
      actor: human
      description: "Name topics, identify mixed or redundant topics, and
        interpret topical evolution using paper-level evidence"

    artifact P_research_foci : pattern(D_topics, D_communities)
      representation form: "named topic descriptions and temporal focus patterns"
      description: "Interpreted research foci characterising communities"

    transform T_assess_topics :
      intent: assess
      manner: "evaluate topic coherence, attribution adequacy, and usefulness"
      input: M_topic_model, V_topics, P_research_foci, F_topic_trends,
             S_topic_current, S_attribution_current, S_quality_current,
             K_provenance_current
      output: K_topic_assessment, S_topic_revised, S_attribution_revised,
              K_provenance_after_topic_assessment
      actor: human
      description: "Assess whether topics are coherent and distinct, whether
        allocation of multi-community papers is misleading, and whether
        topic settings require refinement"

    artifact K_topic_assessment : knowledge(D_topics, M_topic_model)
      representation form: "quality judgment"
      description: "Assessment of topic coherence, coverage, distinction,
        temporal interpretability, and attribution adequacy"

    artifact S_topic_revised : specification
      representation form: "revised topic-model settings and constraints"
      description: "Analyst-guided changes such as merging or splitting
        topics, removing generic terms, seeding concepts, or changing
        topic count"

    artifact S_attribution_revised : specification
      representation form: "revised multi-community allocation rules"
      description: "Justified adjustment of fractional, full-membership, or
        primary-community topic reporting"

    artifact K_provenance_after_topic_assessment : knowledge(
        K_topic_assessment, S_topic_revised, S_attribution_revised)
      representation form: "provenance record"
      description: "Recorded topic edits, attribution decisions, evidence,
        and assessment rationale"

    if K_topic_assessment indicates refinement needed:
      then:
        assign:
          S_topic_current := S_topic_revised
          S_attribution_current := S_attribution_revised
          K_provenance_current := K_provenance_after_topic_assessment
      else:
        assign:
          K_provenance_current := K_provenance_after_topic_assessment
        exit loop L_topic_refinement

end loop L_topic_refinement

# ============================================================================
# LINEAGE MATCHING, EVENT REVIEW, AND LINEAGE-VALIDATION LOOP
# ============================================================================

assign:
  S_lineage_current := S_lineage_matching

loop L_lineage_validation:
  purpose: "Derive and validate community lineages and evolution events
    against membership, structural, topical, paper-level, and citation evidence"
  until: "Lineage links and event labels are sufficiently supported by the
    lineage rules and analyst review, with uncertainty explicitly recorded"
  body:

    transform T_define_lineages :
      intent: define-unit
      manner: "match communities across adjacent temporal windows"
      input: D_communities, F_community_membership, F_author_roles,
             F_topic_mixture, F_topic_similarity, S_lineage_current
      output: D_community_lineages, F_lineage_membership, F_event_candidates
      actor: hybrid
      description: "Match community instances across time, prioritising
        stable-core overlap while considering total membership, topical,
        and structural similarity"

    artifact D_community_lineages : entities
      internal structure: sequence
      embedment: set
      features:
        - id: lifespan
          value structure: atomic
          value type: temporal
          description: "Lineage time span"
      description: "Sequences of related community instances"

    artifact F_lineage_membership : feature(D_communities)
      value structure: relational configuration
      value type: reference
      representation form: "predecessor-successor lineage graph"
      description: "Inferred links between successive community instances"

    artifact F_event_candidates : feature(D_community_lineages)
      value structure: list
      value type: categorical
      description: "Candidate continuation, growth, merge, split, birth,
        dissolution, temporary-disappearance, and reorientation events"

    transform T_characterise_event_evidence :
      intent: characterise
      manner: "compute membership, topical, venue, paper, and citation evidence"
      input: D_authors, D_papers, D_communities, D_community_lineages,
             F_community_membership, F_author_roles, F_topic_mixture,
             F_event_candidates, S_citation_scope
      output: F_overlap, F_member_transitions, F_event_evidence
      actor: machine
      description: "Compute community overlap, mobile and simultaneous
        memberships, shared papers, core changes, topical change, venue
        shifts, and policy-compliant citation evidence"

    artifact F_overlap : feature(D_communities)
      value structure: matrix
      value type: numeric
      representation form: "time-resolved pairwise overlap matrix"
      description: "Shared membership between communities through time"

    artifact F_member_transitions : feature(D_authors)
      value structure: relational configuration
      value type: reference
      representation form: "author-to-community membership histories"
      description: "Researcher mobility and multiple-community participation"

    artifact F_event_evidence : feature(D_community_lineages)
      value structure: vector
      value type: {numeric, categorical, temporal, text}
      description: "Structural, membership, topic, paper, venue, and
        citation evidence for candidate events"

    artifact D_lineage_timeline : entities
      origin: given
      internal structure: elementary
      embedment: time
      features:
        - id: date
          value structure: atomic
          value type: temporal
          description: "Time position"
      description: "Temporal context for lineage displays"

    transform T_contextualise_lineages :
      intent: contextualise
      manner: "time-based lineage arrangement"
      input: D_communities, D_lineage_timeline, F_lineage_membership
      output: A_lineage_timeline
      actor: machine
      description: "Arrange community instances by time and lineage relations"

    artifact A_lineage_timeline : arrangement(D_communities)
      context: D_lineage_timeline
      principle: "temporal placement with lineage-preserving band positions"
      description: "Time-based arrangement of community instances"

    transform T_visualise_evolution :
      intent: visualise
      manner: "coordinated lineage, overlap, and evidence views"
      input: A_lineage_timeline, D_community_lineages, F_event_candidates,
             F_event_evidence, F_overlap, F_member_transitions,
             F_author_roles, F_topic_trends
      output: V_community_evolution
      actor: machine
      description: "Show lineage bands, membership ribbons, overlap matrix,
        selected-community network matrices, topic timelines, and linked
        evidence from papers and citations"

    artifact V_community_evolution : visualisation(A_lineage_timeline,
        D_community_lineages, F_event_candidates, F_event_evidence,
        F_overlap, F_member_transitions, F_author_roles, F_topic_trends)
      layout: "lineage bands with coordinated overlap, member, topic, and
        evidence-detail views"
      form: "alluvial bands, ribbons, heatmaps, matrices, node-link views,
        timelines, and document cards"
      encoding: "band width encodes activity; links encode lineage relation;
        colour identifies communities or topics; highlighting exposes
        members and supporting evidence"
      description: "Workspace for inspecting community evolution"

    transform T_review_evolution_events :
      intent: abstract
      manner: "perception-and-interpretation of evolution events"
      input: V_community_evolution, F_event_candidates, F_event_evidence,
             P_research_foci, K_domain_knowledge
      output: P_reviewed_evolution_events
      actor: human
      description: "Review candidate events and distinguish likely genuine
        structural change from sparse-data, windowing, or matching artifacts"

    artifact P_reviewed_evolution_events : pattern(D_community_lineages)
      representation form: "reviewed event annotations"
      description: "Analyst-reviewed community continuations, merges,
        splits, births, dissolutions, disappearances, and reorientations"

    transform T_assess_lineages :
      intent: assess
      manner: "validate lineage matching and event labels"
      input: D_community_lineages, F_lineage_membership, F_event_candidates,
             F_event_evidence, P_reviewed_evolution_events,
             S_lineage_current, S_quality_current, K_provenance_current
      output: K_lineage_assessment, S_lineage_revised,
              K_provenance_after_lineage_assessment
      actor: hybrid
      description: "Assess whether matching rules and event thresholds
        produce defensible lineages; identify uncertain events and decide
        whether matching policy should be revised"

    artifact K_lineage_assessment : knowledge(D_community_lineages,
        F_event_candidates, F_event_evidence)
      representation form: "quality judgment and uncertainty annotations"
      description: "Assessment of lineage validity, event confidence, and
        need for matching-rule revision"

    artifact S_lineage_revised : specification
      representation form: "revised matching rules and thresholds"
      description: "Justified revisions to core-overlap weighting, topical
        weighting, sparse-period treatment, event thresholds, or temporary-
        disappearance rules"

    artifact K_provenance_after_lineage_assessment : knowledge(
        K_lineage_assessment, S_lineage_revised)
      representation form: "provenance record"
      description: "Recorded lineage settings, event evidence, uncertainty,
        and rationale for accepted or revised matching decisions"

    if K_lineage_assessment indicates lineage refinement needed:
      then:
        assign:
          S_lineage_current := S_lineage_revised
          K_provenance_current := K_provenance_after_lineage_assessment
      else:
        assign:
          K_provenance_current := K_provenance_after_lineage_assessment
        exit loop L_lineage_validation

end loop L_lineage_validation

# ============================================================================
# EVIDENCE-BOUNDED EXPLANATION OF EVOLUTION EVENTS
# ============================================================================

transform T_formulate_event_explanations :
  intent: generate-knowledge
  manner: "formulate evidence-bounded explanatory statements"
  input: P_reviewed_evolution_events, P_research_foci, F_event_evidence,
         S_explanation_standard, K_domain_knowledge
  output: K_evolution_explanations
  actor: human
  description: "Formulate descriptions and hypotheses about why events
    occurred; classify each statement by the available evidence level and
    avoid unsupported causal claims"

artifact K_evolution_explanations : knowledge(P_reviewed_evolution_events,
    P_research_foci, F_event_evidence)
  representation form: "annotated explanatory statements"
  description: "Evidence-bounded interpretations of community evolution,
    labelled as descriptive observation, evidence-backed hypothesis, or
    externally corroborated explanation when external evidence exists"

# ============================================================================
# ROBUSTNESS, RESAMPLING, AND UNCERTAINTY ANALYSIS
# ============================================================================

assign:
  S_robustness_current := S_robustness_analysis

loop L_robustness_analysis:
  purpose: "Test whether substantive community, topic, lineage, and event
    findings persist under plausible alternative analytical choices"
  until: "The analyst has examined the specified sensitivity and resampling
    cases and can classify findings as robust, tentative, or uncertain"
  body:

    transform T_define_alternative_snapshots :
      intent: define-unit
      manner: "alternative time partitioning and resampling"
      input: D_papers, D_authors, F_coauthorship, S_robustness_current
      output: D_alternative_network_snapshots
      actor: machine
      description: "Create resampled or alternative-window collaboration
        snapshots according to the robustness specification"

    artifact D_alternative_network_snapshots : entities
      internal structure: formation
      embedment: time
      description: "Alternative or resampled temporal co-authorship snapshots"

    transform T_fit_alternative_community_model :
      intent: build-model
      manner: "fit alternative overlapping community model"
      input: D_alternative_network_snapshots, F_coauthorship,
             S_robustness_current
      output: M_alternative_community_model
      actor: machine
      description: "Fit community models under alternative resolutions,
        weights, filtering, window, and resampling choices"

    artifact M_alternative_community_model : model(
        D_alternative_network_snapshots, F_coauthorship)
      model type: "overlapping community-detection model"
      description: "Alternative community solution"

    transform T_define_alternative_communities :
      intent: define-unit
      manner: "derive alternative overlapping groups"
      input: D_alternative_network_snapshots, D_authors,
             M_alternative_community_model
      output: D_alternative_communities, F_alternative_membership
      actor: machine
      description: "Define alternative community memberships"

    artifact D_alternative_communities : entities
      internal structure: group/cluster
      embedment: time
      description: "Community instances from an alternative configuration"

    artifact F_alternative_membership : feature(D_authors)
      value structure: matrix
      value type: numeric
      description: "Alternative author-community membership weights"

    transform T_characterise_alternative_communities :
      intent: characterise
      manner: "compute alternative roles and topic corpora"
      input: D_alternative_communities, D_authors, F_alternative_membership,
             F_coauthorship, D_papers, F_text_representation,
             S_role_current, S_robustness_current
      output: F_alternative_roles, F_alternative_corpora
      actor: machine
      description: "Compute comparable role measures and alternative corpora,
        including the selected alternative attribution rule"

    artifact F_alternative_roles : feature(D_authors)
      value structure: vector
      value type: {categorical, numeric, temporal}
      description: "Roles computed for alternative community memberships"

    artifact F_alternative_corpora : feature(D_alternative_communities)
      value structure: matrix
      value type: numeric
      description: "Alternative community document corpora"

    transform T_fit_alternative_topic_model :
      intent: build-model
      manner: "fit alternative topic model"
      input: F_alternative_corpora, S_robustness_current
      output: M_alternative_topic_model
      actor: machine
      description: "Fit topic model under alternative preprocessing,
        topic-count, and attribution settings"

    artifact M_alternative_topic_model : model(F_alternative_corpora)
      model type: "topic model"
      description: "Alternative topical representation"

    transform T_define_alternative_topics :
      intent: define-unit
      manner: "derive alternative topic groups"
      input: M_alternative_topic_model, F_alternative_corpora
      output: D_alternative_topics, F_alternative_topic_mixture
      actor: machine
      description: "Define alternative research topics and mixtures"

    artifact D_alternative_topics : entities
      internal structure: group/cluster
      embedment: set
      description: "Topics from an alternative configuration"

    artifact F_alternative_topic_mixture : feature(D_alternative_communities)
      value structure: vector
      value type: numeric
      description: "Alternative community topic profiles"

    transform T_define_alternative_lineages :
      intent: define-unit
      manner: "match alternative communities across time"
      input: D_alternative_communities, F_alternative_membership,
             F_alternative_roles, F_alternative_topic_mixture,
             S_lineage_current
      output: D_alternative_lineages, F_alternative_events
      actor: machine
      description: "Derive alternative lineage and event structures"

    artifact D_alternative_lineages : entities
      internal structure: sequence
      embedment: set
      description: "Lineages from an alternative configuration"

    artifact F_alternative_events : feature(D_alternative_lineages)
      value structure: list
      value type: categorical
      description: "Alternative lineage-event labels"

    transform T_characterise_robustness :
      intent: characterise
      manner: "measure agreement and confidence across configurations"
      input: D_communities, D_community_lineages, F_community_membership,
             F_author_roles, F_topic_mixture, F_event_candidates,
             D_alternative_communities, D_alternative_lineages,
             F_alternative_membership, F_alternative_roles,
             F_alternative_topic_mixture, F_alternative_events,
             S_robustness_current
      output: F_result_robustness
      actor: machine
      description: "Compute agreement for memberships, stable cores, leaders,
        topics, lineages, and events across alternative and resampled runs"

    artifact F_result_robustness : feature(D_community_lineages)
      value structure: vector
      value type: {numeric, categorical}
      description: "Agreement scores, confidence intervals where applicable,
        and robust, tentative, or uncertain labels"

    transform T_visualise_robustness :
      intent: visualise
      manner: "comparative sensitivity display"
      input: F_result_robustness, D_communities, D_alternative_communities,
             D_community_lineages, D_alternative_lineages
      output: V_robustness
      actor: machine
      description: "Compare primary and alternative solutions and expose
        parameter-sensitive communities, topics, and events"

    artifact V_robustness : visualisation(F_result_robustness, D_communities,
        D_alternative_communities, D_community_lineages,
        D_alternative_lineages)
      layout: "aligned solution comparison and stability summaries"
      form: "matched lineage bands, agreement matrices, and stability plots"
      encoding: "alignment links comparable results; colour and opacity
        encode robustness category and agreement score"
      description: "Visual evidence for uncertainty assessment"

    transform T_assess_robustness :
      intent: assess
      manner: "judge sensitivity-analysis sufficiency"
      input: V_robustness, F_result_robustness, S_robustness_current,
             K_domain_knowledge, K_provenance_current
      output: K_robustness_assessment, S_robustness_revised,
              K_provenance_after_robustness_assessment
      actor: human
      description: "Judge whether examined alternatives are sufficient and
        classify final findings by robustness and uncertainty"

    artifact K_robustness_assessment : knowledge(F_result_robustness)
      representation form: "quality judgment and uncertainty classification"
      description: "Assessment of robustness coverage and confidence in
        communities, roles, topics, lineages, and events"

    artifact S_robustness_revised : specification
      representation form: "revised resampling and sensitivity settings"
      description: "Additional justified alternatives or resampling cases
        required for adequate uncertainty assessment"

    artifact K_provenance_after_robustness_assessment : knowledge(
        K_robustness_assessment, S_robustness_revised)
      representation form: "provenance record"
      description: "Recorded sensitivity cases, outcomes, uncertainty
        decisions, and rationale"

    if K_robustness_assessment indicates additional testing needed:
      then:
        assign:
          S_robustness_current := S_robustness_revised
          K_provenance_current := K_provenance_after_robustness_assessment
      else:
        assign:
          K_provenance_current := K_provenance_after_robustness_assessment
        exit loop L_robustness_analysis

end loop L_robustness_analysis

# ============================================================================
# SYNTHESIS AND REPORTING
# ============================================================================

transform T_synthesise_findings :
  intent: generate-knowledge
  manner: "formulate evidence-linked analytical report"
  input: D_community_lineages, F_author_roles, F_community_metrics,
         P_research_foci, P_reviewed_evolution_events,
         K_evolution_explanations, F_result_robustness,
         K_robustness_assessment, K_provenance_current,
         S_cleaning_policy, S_citation_scope
  output: K_community_findings
  actor: human
  description: "Report active communities and long-term lineages, stable
    cores and leaders, research foci, membership overlap and mobility,
    reviewed evolution events, evidential status of explanations, and
    uncertainty"

artifact K_community_findings : knowledge(D_community_lineages,
    P_research_foci, P_reviewed_evolution_events, F_result_robustness)
  representation form: "annotated report and reproducibility package"
  description: "Final evidence-linked account of the evolving intellectual
    structure of the IEEE VIS research community, including methodological
    provenance and uncertainty qualifications"

Changes relative to the previous ATWL workflow

Implemented as recommended:

• Added explicit quality criteria, role-definition rules, paper-attribution rules, citation scope, robustness settings, and an evidence hierarchy as specifications.
• Added a formal data-cleaning and citation-semantics policy after the corpus audit.
• Made stable-core, leader, peripheral, and boundary-spanner definitions explicitly controlled by S_role_definition.
• Added lightweight topical screening inside the community-refinement loop, so community acceptance considers intellectual coherence as well as network structure.
• Added a distinct lineage-validation loop, allowing analyst review to revise event thresholds and matching weights.
• Expanded robustness analysis to specify alternative time windows, model choices, attribution rules, and resampling/agreement procedures.
• Added evolving provenance records after audit, community assessment, topic assessment, lineage validation, and robustness assessment.
• Added an explicit evidence-bounded explanation phase distinguishing observations, hypotheses, and externally corroborated explanations.

Implemented differently from the earlier recommendation:

• Rather than fully combining detailed topic modelling and community detection in one joint loop, the revision uses lightweight topical screening during community refinement and retains a separate detailed, human-steered topic loop afterward. This avoids extensive topic editing for unstable candidate communities while still allowing topic coherence to influence community selection.
• Citation analysis is controlled by S_citation_scope rather than assuming citation links are available. Citation counts remain impact indicators; citation-flow analysis is only performed when links are available.
• External corroboration is represented as a possible evidence level, not as a required external-data phase, because no such data source was included in the task.

No recommended modification was rejected. The provenance and specification additions were retained because they correspond to real analytical decisions that materially affect reproducibility and interpretation.