workflow eventthread-sequence-summarization template: define-unit (filter) -> characterise (align) -> define-unit (segment stages) -> characterise (encode tensor) -> build-model -> define-unit (derive threads) -> contextualise (layout) -> loop(define-unit (cluster-by-similarity) -> visualise -> abstract -> assess) -> generate-knowledge description: "Visual analytics workflow for summarizing large collections of heterogeneous event sequences into latent threads via tensor decomposition, laying out threads to reveal their stage-wise evolution, and interactively clustering threads into latent stage categories to support interpretation of progression patterns" # ============================================================================ # INPUT # ============================================================================ artifact D_sequences : entities origin: given internal structure: sequence embedment: set features: - id: f_event_types value structure: list value type: categorical description: "Ordered event types within each sequence" - id: f_event_times value structure: list value type: temporal description: "Occurrence times of events within each sequence" description: "Heterogeneous timestamped event sequences, one per analytic entity, containing variable event types and lengths" # ============================================================================ # STEP 1: FILTER NOISY EVENTS # ============================================================================ transform T_filter : intent: define-unit manner: "filter low-importance events using TF-IDF-based scoring" input: D_sequences output: D_sequences_filtered actor: machine description: "Remove events with low TF-IDF importance, retaining characteristic events that differentiate each sequence from the collection" artifact D_sequences_filtered : entities internal structure: sequence embedment: set description: "Denoised event sequences retaining characteristic events after removing low-importance events for each entity" # ============================================================================ # STEP 2: ALIGN SEQUENCES TO A COMMON TIME ORIGIN # ============================================================================ transform T_align : intent: characterise manner: "shift events to a common relative time origin" input: D_sequences_filtered output: F_relative_time actor: machine description: "Compute each event's temporal position relative to a shared reference point, replacing variable absolute timestamps with comparable relative times" artifact F_relative_time : feature(D_sequences_filtered) value structure: vector value type: temporal description: "Relative time position of each retained event with respect to a common origin for its sequence" # ============================================================================ # SPECIFICATION FOR STAGE SEGMENTATION AND DECOMPOSITION # ============================================================================ artifact S_stage_spec : specification origin: given representation form: "parameter settings" description: "Analyst-defined stage segmentation granularity (fixed duration or event-count boundaries) and number of latent threads for tensor decomposition" # ============================================================================ # STEP 3: SEGMENT INTO SHARED STAGES # ============================================================================ transform T_segment : intent: define-unit manner: "time-partitioning into discrete stages" input: D_sequences_filtered, F_relative_time, S_stage_spec output: D_stages actor: machine description: "Define a shared ordered sequence of discrete stages used to fold events across all entity sequences" artifact D_stages : entities internal structure: elementary embedment: sequence description: "Ordered discrete stages representing shared temporal units of progression across all sequences (e.g., calendar years, maintenance rounds)" # ============================================================================ # STEP 4: ENCODE AS EVENT-STAGE-ENTITY TENSOR # ============================================================================ transform T_encode : intent: characterise manner: "fold events into stages and encode as event-stage-entity tensor" input: D_sequences_filtered, F_relative_time, D_stages output: F_tensor actor: machine description: "Aggregate events within each stage per entity and represent the result as a multi-way tensor of event, stage, and entity co-occurrence" artifact F_tensor : feature(D_sequences_filtered) value structure: relational configuration value type: numeric representation form: "three-way event-stage-entity incidence tensor with non-negative occurrence counts" description: "Multi-way representation of event occurrences across event types, shared stages, and entities" # ============================================================================ # STEP 5: DECOMPOSE TENSOR TO EXTRACT LATENT THREADS # ============================================================================ transform T_decompose : intent: build-model manner: "CP tensor decomposition with an orthogonality constraint on the stage factor" input: F_tensor, S_stage_spec output: M_tensor_model actor: machine description: "Factorise the event-stage-entity tensor into component weights and event, stage, and entity factor matrices, constraining the stage factor to support interpretable thread evolution" artifact M_tensor_model : model(F_tensor, S_stage_spec) model type: "tensor decomposition model" representation form: "CP decomposition with component weights and event, stage, and entity factor matrices" description: "Factorised representation from which latent threads and their event, stage, and entity distributions are derived" # ============================================================================ # STEP 6: DEFINE LATENT THREADS # ============================================================================ transform T_define_threads : intent: define-unit manner: "derive latent threads from decomposition factor matrices" input: M_tensor_model output: D_threads, F_thread_event_relevance, F_thread_stage_likelihood, F_thread_entity_relevance actor: machine description: "Define threads as latent sequential patterns, each characterised by its relevance to events, likelihood across stages, and relevance to entities" artifact D_threads : entities internal structure: sequence embedment: set description: "Latent threads representing summarized clusters of similar event sequences, each evolving across the shared stage sequence" artifact F_thread_event_relevance : feature(D_threads) value structure: vector value type: numeric description: "Relevance of each event type to each thread" artifact F_thread_stage_likelihood : feature(D_threads) value structure: vector value type: numeric description: "Occurrence likelihood of each thread at each stage" artifact F_thread_entity_relevance : feature(D_threads) value structure: vector value type: numeric description: "Relevance of each entity to each thread" # ============================================================================ # STEP 7: LAYOUT THREADS # ============================================================================ transform T_layout : intent: contextualise manner: "optimisation-based layout balancing thread similarity and crossing minimisation" input: D_threads, F_thread_event_relevance, F_thread_stage_likelihood output: D_layout_space, A_thread_layout actor: machine description: "Compute positions for threads at each stage that group similar threads together while minimising crossings across stages" artifact D_layout_space : entities internal structure: elementary features: - id: f_axes value structure: atomic value type: numeric description: "Number of spatial dimensions (stage axis, vertical position)" description: "Computed two-dimensional layout space for displaying threads across stages" artifact A_thread_layout : arrangement(D_threads) context: D_layout_space principle: "stage-wise vertical positioning balancing thread similarity and crossing minimisation" description: "Positions of threads within each stage reflecting event-based similarity and continuity across stages" # ============================================================================ # SPECIFICATION FOR STAGE-CATEGORY CLUSTERING (initial loop artifact) # ============================================================================ artifact S_clustering : specification origin: given representation form: "parameter settings" description: "Initial distance threshold controlling grouping of threads into latent stage categories at each stage" # ============================================================================ # STEP 8: INTERACTIVE STAGE-CATEGORY CLUSTERING LOOP # ============================================================================ loop L_stage_clustering: purpose: "Iteratively adjust the thread clustering threshold to reveal meaningful and interpretable latent stage categories and their evolution across stages" until: "Latent stage categories and their merging/splitting patterns are interpretable and reveal meaningful evolution structures" body: transform T_cluster_threads : intent: define-unit manner: "cluster-by-similarity (threshold-based grouping within each stage)" input: D_threads, A_thread_layout, S_clustering output: D_stage_categories, F_category_label actor: hybrid description: "Group neighbouring threads within each stage into latent stage categories based on an adjustable distance threshold" artifact D_stage_categories : entities internal structure: group/cluster embedment: {set, time} description: "Groups of threads with similar event profiles within a given stage, representing latent stage categories" artifact F_category_label : feature(D_threads) value structure: vector value type: categorical description: "Thread's latent stage category membership at each stage" transform T_visualise_categories : intent: visualise manner: "line-map thread view with coordinated contextual panels" input: A_thread_layout, F_category_label, F_thread_event_relevance, F_thread_stage_likelihood, F_thread_entity_relevance, D_threads, D_sequences_filtered, D_stages output: V_threads actor: machine description: "Render threads as stage-segmented stripes grouped into latent stage categories, with event and entity context in coordinated views for inspection" artifact V_threads : visualisation(A_thread_layout, F_category_label, F_thread_event_relevance, F_thread_stage_likelihood, F_thread_entity_relevance, D_sequences_filtered, D_stages) layout: "line-map with threads divided into stage segments; coordinated contextual panels" form: "coloured linear stripes with event nodes, optional entity-volume backgrounds, and contextual list and flow views" encoding: "stage and vertical position from A_thread_layout; category grouping from F_category_label; stripe opacity and dashed segments from F_thread_stage_likelihood; event-node size from event relevance weighted by stage likelihood; event-node colour from event type; background height from entity relevance weighted by stage likelihood; contextual views show selected raw events and entities" description: "Coordinated multi-view visualisation of latent threads, latent stage categories, and supporting event-sequence context" transform T_interpret : intent: abstract manner: "perception and interpretation of thread/category merging, splitting, and event composition" input: V_threads, D_stage_categories output: P_patterns actor: human description: "Interpret the semantic meaning of latent stage categories and their evolution (merges, splits, branches) across stages" artifact P_patterns : pattern(D_stage_categories, D_threads) representation form: "textual labels describing category composition and evolution" description: "Interpreted descriptions of latent stage categories, their characteristic events and entities, and how they merge, split, or diverge across stages" transform T_assess_clustering : intent: assess manner: "evaluate interpretability and coherence of resulting stage categories" input: V_threads, P_patterns output: K_cluster_quality actor: human description: "Assess whether the current clustering threshold yields coherent, interpretable stage categories and meaningful evolution patterns" artifact K_cluster_quality : knowledge(D_stage_categories) representation form: "quality judgment" description: "Assessment of stage-category coherence, interpretability, and whether further threshold adjustment is needed" if K_cluster_quality indicates satisfactory clustering: then: exit loop L_stage_clustering else: transform T_adjust_threshold : intent: generate-knowledge manner: "adjust clustering distance threshold based on assessment" input: K_cluster_quality, V_threads, S_clustering output: S_clustering_r actor: human description: "Adjust the distance threshold to merge or separate thread groupings and better reveal meaningful stage categories" artifact S_clustering_r : specification representation form: "parameter settings" description: "Updated distance threshold after analyst refinement" assign: S_clustering := S_clustering_r end loop L_stage_clustering # ============================================================================ # STEP 9: SYNTHESISE FINDINGS # ============================================================================ transform T_synthesise : intent: generate-knowledge manner: "formulate statements about domain progression patterns" input: P_patterns, K_cluster_quality output: K_findings actor: human description: "Synthesise insights about typical and divergent progression pathways, their distinguishing event characteristics, and their evolution across stages" artifact K_findings : knowledge(P_patterns, K_cluster_quality) representation form: "statements and explanations" description: "Documented domain insights about typical progression pathways, divergent subgroup behaviours, and characteristics revealed by thread evolution across stages"