{
  "chat_uuid": "2dca87e1-a62f-4248-8bad-0bed0b1b49e9",
  "history": [
    {
      "content": "You have received two files: the ATWL specification, which defines a formal language for representing visual analytics workflows, and an example workflow expressed in ATWL. Please read both carefully before we proceed.\n\nThe ATWL specification describes the language's artifact types, transform intents, actor designations, loop constructs, and syntax. The example workflow illustrates how these elements are used together to describe a specific visual analytics process. Note that the example represents one type of workflow among many possible; its specific analytical structure reflects its particular task and should not be treated as a template that must be followed for other tasks. What the example illustrates is how the language is used — the syntax, the vocabulary, and the conventions for organising artifacts, transforms, and loops — not what analytical structure any given task requires.\n\nIn the next message I will describe a specific visual analytics design task. Your job will be to design an analytical workflow that addresses the task, and to express your recommendation both in natural language (for readability) and in ATWL (for structural precision). The natural-language description should explain the analytical logic of the workflow; the ATWL representation should formally specify the artifacts, transforms, loops, actors, and control flow.\n\nBefore we proceed, could you briefly confirm that you have understood the ATWL specification and the example, and let me know if you have any questions about the language or the syntax?",
      "role": "user"
    },
    {
      "content": "I have understood the ATWL specification and the example workflow.\n\nATWL represents visual-analytics workflows through:\n- **Artifacts**: entities, features, arrangements, visualisations, patterns, models, knowledge, and specifications.\n- **Transforms**: each with one generic intent, optional manner, explicit inputs/outputs, and a human, machine, or hybrid actor.\n- **Control flow**: loops, conditionals, and assignments for iterative refinement.\n- **Workflow templates**: concise summaries of the primary analytical stages.\n\nI also understand that the example demonstrates ATWL syntax and conventions—not a required analytical structure. For the next task, I will derive a workflow appropriate to its own analytical goals, explain its logic in natural language, and provide a formally structured ATWL specification.\n\nI have no questions about the language or syntax at this stage.",
      "role": "assistant"
    },
    {
      "content": "A public health team in Vastopolis, a metropolitan area of about two million residents, has observed a sharp increase in flu-like illnesses over the past few days, with several deaths believed to be associated with the current outbreak. The team wants to understand where the outbreak started and how the infection is being transmitted (person-to-person, airborne, waterborne, or something else) so that emergency management resources can be deployed appropriately.\n\nThe available data consists of: geolocated microblog messages posted from mobile devices over a period of several days, each with a timestamp, GPS coordinates, and text content; a map of the metropolitan area showing highways, hospitals, landmarks, and water bodies; population statistics for 13 city zones; and daily weather observations including wind speed and direction.\n\nThe analyst will need to inspect intermediate results visually and use domain knowledge to interpret patterns and make judgements at each stage.\n\nDesign an analytical workflow for this problem.",
      "role": "user"
    },
    {
      "content": "<think>**Building models for hypotheses**\n\nI'm considering how to build models based on different hypotheses. For a person-to-person model, I might look at factors like population density and how locations are connected by roads. Microblogs don't provide much on contacts or mobility, so I’d use a statistical diffusion model focusing on lag proximity and population along highways. For airborne diffusion, I’d consider wind direction and how it affects downwind areas. Waterborne models might rely on proximity and downstream effects.\n\n**Evaluating relevance loops**\n\nI'm considering how to handle a return to the relevance loop or a recompute if the posts aren't adequate. There's talk of ATWL control being nested in an overall loop, but that might be overkill. It seems the requirement is for an analyst to inspect intermediate stages, rather than formalizing every possible backtracking. I also see that there's a visual relevance and visual hypothesis loop in play here that needs attention.\n\n**Considering model artifact updates**\n\nI'm thinking about the potential need for model artifact versions within the loop for `M_competing`. It seems like using the same reference rather than just assigning values could be useful. Maybe I should take the initial output of `M_classifier` and then use the refined version from `M_r`. As I loop through, I could fit the latest model with the most updated `S` in each iteration. That could help streamline the process!\n\n**Planning design output**\n\nThe question is about designing a recommendation for \"where the outbreak started and how it was transmitted.\" I have a candidate source and a clear mechanism in my posterior rankings, which is a good start! I think I might need around 400 lines for the output. However, I want to aim for clarity and brevity, so I’ll try to condense it down to about 230 lines. I’ll use code-fenced text with the `atwl` language identifier, avoiding any math or boxed elements.</think>\n\n## Analytical logic\n\nThis workflow treats microblogs as a **syndromic proxy**, not as confirmed infection records. It first identifies messages likely to describe flu-like illness, then aggregates them into zone-by-day outbreak indicators normalized by population.\n\nThe analyst visually examines maps and time series to identify **candidate origin areas** based on the earliest *sustained excess* of illness-related activity—not simply the first isolated post. The workflow then compares alternative transmission explanations:\n\n- **Person-to-person:** spread through adjacent or highly connected zones, population density, and transport corridors.\n- **Airborne:** downwind and time-lagged propagation consistent with observed wind.\n- **Water-associated:** clustering near water bodies or spatial patterns plausibly associated with water infrastructure.\n- **Other/local source:** a localized point-source pattern near a landmark, facility, or event area.\n\nThe models provide comparative evidence rather than causal proof. The analyst iteratively inspects evidence, adjusts assumptions or lags, and may conclude that the available data cannot reliably distinguish transmission routes. The final output identifies candidate source locations, ranks transmission explanations with uncertainty, recommends resource deployment, and specifies additional data required for confirmation—especially confirmed case, death, contact, environmental, and water-quality data.\n\n```atwl\nworkflow vastopolis-flu-outbreak-investigation\n  template: loop(characterise -> visualise -> abstract -> assess) ->\n            define-unit -> define-unit -> characterise ->\n            contextualise -> visualise -> abstract ->\n            loop(build-model -> characterise -> visualise -> abstract -> assess) ->\n            generate-knowledge\n  description: \"Human-in-the-loop investigation of a suspected flu-like\n    outbreak using geolocated microblogs as a syndromic surveillance proxy.\n    The workflow identifies candidate origins and compares person-to-person,\n    airborne, water-associated, and other local-source explanations while\n    preserving uncertainty.\"\n\n# ============================================================================\n# INPUT DATA, CONTEXT, AND PRIOR KNOWLEDGE\n# ============================================================================\n\nartifact D_posts : entities\n  origin: given\n  internal structure: elementary\n  embedment: {space, time}\n  features:\n    - id: timestamp\n      value structure: atomic\n      value type: temporal\n      description: \"Time at which the message was posted\"\n    - id: location\n      value structure: atomic\n      value type: spatial\n      description: \"GPS coordinate associated with the mobile device\"\n    - id: text\n      value structure: atomic\n      value type: text\n      description: \"Microblog message content\"\n  description: \"Geolocated microblog messages posted during the suspected\n    outbreak period\"\n\nartifact D_zones : entities\n  origin: given\n  internal structure: region\n  embedment: space\n  features:\n    - id: zone_id\n      value structure: atomic\n      value type: reference\n      description: \"Identifier of one of the 13 metropolitan zones\"\n    - id: population\n      value structure: atomic\n      value type: numeric\n      description: \"Resident population of the zone\"\n    - id: geometry\n      value structure: atomic\n      value type: spatial\n      description: \"Zone boundary geometry\"\n  description: \"Metropolitan zones used for population-normalised\n    aggregation and emergency-resource planning\"\n\nartifact D_geography : entities\n  origin: given\n  internal structure: elementary\n  embedment: space\n  features:\n    - id: geographic_feature_type\n      value structure: atomic\n      value type: categorical\n      description: \"Highway, hospital, landmark, water body, or other map feature\"\n    - id: geometry\n      value structure: atomic\n      value type: spatial\n      description: \"Spatial geometry of the geographic feature\"\n  description: \"Reference map features, including highways, hospitals,\n    landmarks, and water bodies\"\n\nartifact D_weather : entities\n  origin: given\n  internal structure: elementary\n  embedment: time\n  features:\n    - id: wind_speed\n      value structure: atomic\n      value type: numeric\n      description: \"Observed daily wind speed\"\n    - id: wind_direction\n      value structure: atomic\n      value type: spatial\n      description: \"Observed daily wind direction\"\n  description: \"Daily weather observations during the investigation period\"\n\nartifact K_epidemiology : knowledge(D_posts, D_zones, D_weather)\n  origin: given\n  representation form: \"domain rules and expected signatures\"\n  description: \"Public-health knowledge about flu-like symptom language,\n    plausible reporting biases, plausible transmission signatures, and the\n    distinction between syndromic signals and confirmed infections\"\n\nartifact S_relevance : specification\n  origin: given\n  representation form: \"method choice and thresholds\"\n  description: \"Initial rules and model settings for identifying posts likely\n    to describe flu-like illness, including symptom terms, negation handling,\n    language variants, confidence threshold, and exclusion criteria\"\n\nartifact S_hypothesis : specification\n  origin: given\n  representation form: \"comparative modelling configuration\"\n  description: \"Initial temporal lags, spatial neighbourhood definitions,\n    candidate transmission mechanisms, uncertainty treatment, and criteria\n    for comparing person-to-person, airborne, water-associated, and\n    local-source explanations\"\n\n# ============================================================================\n# STEP 1: ITERATIVELY IDENTIFY ILLNESS-RELATED MICROBLOG POSTS\n# ============================================================================\n\nloop L_relevance_refinement:\n  purpose: \"Develop an interpretable and sufficiently reliable syndromic\n    signal from unstructured microblog text\"\n  until: \"The analyst judges that illness-related messages are adequately\n    distinguished from unrelated, historical, sarcastic, and non-local posts\"\n  body:\n\n    transform T_score_posts :\n      intent: characterise\n      manner: \"text encoding and illness relevance scoring\"\n      input: D_posts, K_epidemiology, S_relevance\n      output: F_illness_score\n      actor: hybrid\n      description: \"Extract symptom expressions, illness mentions, negation,\n        severity language, and relevance confidence from post text\"\n\n    artifact F_illness_score : feature(D_posts)\n      value structure: vector\n      value type: {numeric, categorical}\n      representation form: \"illness relevance score and linguistic indicators\"\n      description: \"Per-post probability of describing a current local\n        flu-like illness and associated symptom-language indicators\"\n\n    transform T_visualise_posts :\n      intent: visualise\n      manner: \"linked text and map inspection\"\n      input: D_posts, F_illness_score\n      output: V_post_relevance\n      actor: machine\n      description: \"Show sampled messages, relevance scores, locations, and\n        posting times so the analyst can inspect false positives and omissions\"\n\n    artifact V_post_relevance : visualisation(D_posts, F_illness_score)\n      layout: \"linked message list, map, and time histogram\"\n      form: \"text records, coloured map points, and bars\"\n      encoding: \"colour indicates illness relevance score; map position comes\n        from GPS coordinates; time histogram shows posting volume\"\n      description: \"Interactive inspection view for validating the syndromic\n        message-identification process\"\n\n    transform T_interpret_language :\n      intent: abstract\n      manner: \"perception-and-interpretation\"\n      input: V_post_relevance, F_illness_score, K_epidemiology\n      output: P_language_patterns\n      actor: human\n      description: \"Identify meaningful symptom vocabulary, ambiguous phrases,\n        reporting artefacts, and systematic false-positive message types\"\n\n    artifact P_language_patterns : pattern(D_posts, F_illness_score)\n      representation form: \"annotated linguistic categories\"\n      description: \"Observed illness-language patterns and misleading message\n        categories relevant to flu-like syndromic surveillance\"\n\n    transform T_assess_relevance :\n      intent: assess\n      manner: \"judge classification adequacy\"\n      input: V_post_relevance, P_language_patterns, F_illness_score\n      output: K_relevance_quality, S_relevance_r\n      actor: human\n      description: \"Assess whether the relevance signal is sufficiently\n        credible for aggregate surveillance and, if needed, formulate\n        revised scoring rules or thresholds\"\n\n    artifact K_relevance_quality : knowledge(F_illness_score, V_post_relevance)\n      representation form: \"quality judgment\"\n      description: \"Analyst judgment about the adequacy, bias, and limitations\n        of the illness-message relevance classification\"\n\n    artifact S_relevance_r : specification\n      representation form: \"revised method choice and thresholds\"\n      description: \"Updated linguistic rules, confidence threshold, or\n        exclusion criteria for illness-message identification\"\n\n    if K_relevance_quality indicates refinement needed:\n      then:\n        assign:\n          S_relevance := S_relevance_r\n      else:\n        exit loop L_relevance_refinement\n\nend loop L_relevance_refinement\n\n# ============================================================================\n# STEP 2: DEFINE OUTBREAK UNITS AND COMPUTE SPATIOTEMPORAL INDICATORS\n# ============================================================================\n\ntransform T_select_illness_posts :\n  intent: define-unit\n  manner: \"filter by illness relevance\"\n  input: D_posts, F_illness_score, S_relevance\n  output: D_illness_posts\n  actor: hybrid\n  description: \"Retain posts judged sufficiently likely to describe a\n    current flu-like illness\"\n\nartifact D_illness_posts : entities\n  internal structure: elementary\n  embedment: {space, time}\n  features:\n    - id: illness_confidence\n      value structure: atomic\n      value type: numeric\n      description: \"Confidence that the post describes flu-like illness\"\n    - id: symptom_terms\n      value structure: list\n      value type: text\n      description: \"Extracted symptom and illness expressions\"\n  description: \"Filtered geolocated posts forming the syndromic surveillance\n    signal for the outbreak investigation\"\n\ntransform T_aggregate_zone_days :\n  intent: define-unit\n  manner: \"space-time aggregation by zone and day\"\n  input: D_illness_posts, D_zones\n  output: D_zone_days\n  actor: machine\n  description: \"Aggregate illness-related posts into daily episodes for each\n    city zone, retaining constituent messages for drill-down inspection\"\n\nartifact D_zone_days : entities\n  internal structure: episode\n  embedment: {space, time}\n  features:\n    - id: zone_id\n      value structure: atomic\n      value type: reference\n      description: \"Zone represented by the zone-day episode\"\n    - id: date\n      value structure: atomic\n      value type: temporal\n      description: \"Calendar day represented by the episode\"\n  description: \"Zone-by-day syndromic surveillance episodes containing\n    illness-related microblog messages\"\n\ntransform T_characterise_outbreak :\n  intent: characterise\n  manner: \"population-normalised outbreak profiling\"\n  input: D_zone_days, D_zones\n  output: F_outbreak_signal\n  actor: machine\n  description: \"Compute illness-post counts, population-normalised rates,\n    temporal changes, excess activity, and uncertainty caused by low volume\"\n\nartifact F_outbreak_signal : feature(D_zone_days)\n  value structure: vector\n  value type: numeric\n  representation form: \"daily syndromic incidence profile\"\n  description: \"Counts, rates per population, growth measures, anomaly scores,\n    and uncertainty indicators for each zone-day\"\n\ntransform T_characterise_context :\n  intent: characterise\n  manner: \"derive environmental and propagation relations\"\n  input: D_zone_days, D_zones, D_geography, D_weather\n  output: F_exposure_context, F_propagation_relations\n  actor: machine\n  description: \"Derive zone-day descriptors and lagged relations involving\n    population density, spatial adjacency, highways, hospitals, water\n    proximity, wind direction, and wind speed\"\n\nartifact F_exposure_context : feature(D_zone_days)\n  value structure: vector\n  value type: {numeric, categorical}\n  description: \"Population, geographic, weather, and infrastructure context\n    associated with each zone-day\"\n\nartifact F_propagation_relations : feature(D_zone_days)\n  value structure: relational configuration\n  representation form: \"directed lagged zone-day relation graph\"\n  description: \"Candidate propagation relations between zone-days based on\n    adjacency, transport connectivity, wind alignment, and water proximity\"\n\n# ============================================================================\n# STEP 3: VISUALLY IDENTIFY CANDIDATE ORIGIN AREAS AND DIFFUSION PATTERNS\n# ============================================================================\n\ntransform T_arrange_zone_days :\n  intent: contextualise\n  manner: \"map-based\"\n  input: D_zone_days, D_zones\n  output: A_zone_map\n  actor: machine\n  description: \"Place each zone-day episode in its corresponding metropolitan\n    zone so time-filtered outbreak maps can be inspected\"\n\nartifact A_zone_map : arrangement(D_zone_days)\n  context: D_zones\n  principle: \"zone identifier mapped to zone geometry\"\n  description: \"Map arrangement of zone-day outbreak episodes\"\n\ntransform T_visualise_situation :\n  intent: visualise\n  manner: \"coordinated spatiotemporal outbreak display\"\n  input: A_zone_map, F_outbreak_signal, F_exposure_context,\n         D_geography, D_weather\n  output: V_outbreak_situation\n  actor: machine\n  description: \"Display animated and time-linked maps, zone time series,\n    wind vectors, population-normalised rates, highways, hospitals, and\n    water bodies\"\n\nartifact V_outbreak_situation : visualisation(A_zone_map, F_outbreak_signal,\n    F_exposure_context, D_geography, D_weather)\n  layout: \"linked metropolitan map, zone time series, and temporal overview\"\n  form: \"choropleth regions, proportional symbols, line graphs, and wind vectors\"\n  encoding: \"zone colour and symbol size show population-normalised illness\n    activity; time controls select dates; overlays show geographic features\n    and observed wind direction\"\n  description: \"Interactive situational-awareness display for locating early\n    sustained activity and visually inspecting possible spread pathways\"\n\ntransform T_abstract_observed_patterns :\n  intent: abstract\n  manner: \"perception-and-interpretation\"\n  input: V_outbreak_situation, F_outbreak_signal, F_propagation_relations\n  output: P_observed_diffusion\n  actor: human\n  description: \"Identify candidate source zones, timing of sustained excess,\n    wave-like spread, spatial clusters, anomalies, and patterns inconsistent\n    with a single simple outbreak explanation\"\n\nartifact P_observed_diffusion : pattern(D_zone_days, F_outbreak_signal)\n  representation form: \"candidate foci and spatiotemporal pattern descriptions\"\n  description: \"Analyst-identified candidate origin areas and observed\n    diffusion structures, explicitly distinguished from confirmed sources\"\n\n# ============================================================================\n# STEP 4: ITERATIVELY COMPARE TRANSMISSION EXPLANATIONS\n# ============================================================================\n\nloop L_hypothesis_comparison:\n  purpose: \"Compare plausible transmission mechanisms and determine whether\n    the available evidence supports a defensible operational conclusion\"\n  until: \"A best-supported explanation, or a justified conclusion that the\n    data cannot distinguish mechanisms, has been reached with documented\n    uncertainty\"\n  body:\n\n    transform T_build_mechanism_models :\n      intent: build-model\n      manner: \"fit competing spatiotemporal transmission models\"\n      input: F_outbreak_signal, F_exposure_context, F_propagation_relations,\n             P_observed_diffusion, K_epidemiology, S_hypothesis\n      output: M_mechanism_models\n      actor: hybrid\n      description: \"Fit and compare models representing person-to-person\n        diffusion, wind-associated airborne spread, water-associated\n        clustering, and localized or other point-source mechanisms\"\n\n    artifact M_mechanism_models : model(F_outbreak_signal,\n        F_exposure_context, F_propagation_relations)\n      model type: \"comparative spatiotemporal explanatory models\"\n      representation form: \"set of competing lagged diffusion and source models\"\n      description: \"Alternative formal explanations of observed zone-day\n        illness activity; model comparison measures evidence rather than\n        proving a causal transmission route\"\n\n    transform T_characterise_model_evidence :\n      intent: characterise\n      manner: \"compute model fit and diagnostic evidence\"\n      input: D_zone_days, F_outbreak_signal, F_exposure_context,\n             F_propagation_relations, M_mechanism_models\n      output: F_hypothesis_evidence\n      actor: machine\n      description: \"Calculate comparative fit, residuals, uncertainty,\n        predicted propagation, and local evidence for each transmission\n        hypothesis\"\n\n    artifact F_hypothesis_evidence : feature(D_zone_days)\n      value structure: vector\n      value type: numeric\n      representation form: \"hypothesis fit, residual, and uncertainty profiles\"\n      description: \"Zone-day evidence indicating how well competing\n        explanations account for observed syndromic activity\"\n\n    transform T_visualise_hypothesis_evidence :\n      intent: visualise\n      manner: \"comparative model diagnostics display\"\n      input: A_zone_map, F_outbreak_signal, F_hypothesis_evidence,\n             M_mechanism_models, D_geography, D_weather\n      output: V_hypothesis_evidence\n      actor: machine\n      description: \"Compare observed and predicted spread, residual maps,\n        temporal lags, downwind alignment, water proximity, and transport\n        connectivity in coordinated views\"\n\n    artifact V_hypothesis_evidence : visualisation(A_zone_map,\n        F_outbreak_signal, F_hypothesis_evidence, M_mechanism_models,\n        D_geography, D_weather)\n      layout: \"linked observed-versus-predicted maps, timelines, and\n        hypothesis comparison panels\"\n      form: \"choropleths, flow arrows, residual marks, line graphs, and\n        comparative score plots\"\n      encoding: \"map panels show observed and model-predicted zone activity;\n        arrows show candidate propagation relations; colour and residual\n        marks show local model agreement or disagreement\"\n      description: \"Interactive evidence display used to inspect whether\n        apparent transmission signatures are coherent, robust, and plausible\"\n\n    transform T_interpret_mechanism_evidence :\n      intent: abstract\n      manner: \"perception-and-interpretation\"\n      input: V_hypothesis_evidence, F_hypothesis_evidence,\n             K_epidemiology\n      output: P_mechanism_evidence\n      actor: human\n      description: \"Interpret model evidence in public-health context;\n        identify support, contradictions, confounding, and uncertainty for\n        each candidate transmission mechanism\"\n\n    artifact P_mechanism_evidence : pattern(M_mechanism_models,\n        F_hypothesis_evidence)\n      representation form: \"ranked mechanism evidence statements\"\n      description: \"Interpreted regularities supporting or weakening\n        person-to-person, airborne, water-associated, or local-source\n        explanations\"\n\n    transform T_assess_hypotheses :\n      intent: assess\n      manner: \"evaluate explanatory adequacy and operational usefulness\"\n      input: V_hypothesis_evidence, P_mechanism_evidence,\n             M_mechanism_models, K_epidemiology\n      output: K_hypothesis_assessment, S_hypothesis_r\n      actor: human\n      description: \"Judge whether model assumptions, evidence, uncertainty,\n        and visual explanations are adequate for a response decision;\n        formulate revised lags, constraints, or comparison settings if needed\"\n\n    artifact K_hypothesis_assessment : knowledge(M_mechanism_models,\n        P_mechanism_evidence)\n      representation form: \"quality judgment and uncertainty assessment\"\n      description: \"Assessment of whether the evidence adequately distinguishes\n        mechanisms, identifies a plausible candidate source, and supports\n        emergency action\"\n\n    artifact S_hypothesis_r : specification\n      representation form: \"revised comparative modelling configuration\"\n      description: \"Updated temporal lags, spatial relations, constraints,\n        or robustness checks required for another hypothesis-comparison cycle\"\n\n    if K_hypothesis_assessment indicates further comparison needed:\n      then:\n        assign:\n          S_hypothesis := S_hypothesis_r\n      else:\n        exit loop L_hypothesis_comparison\n\nend loop L_hypothesis_comparison\n\n# ============================================================================\n# STEP 5: PRODUCE OPERATIONAL FINDINGS AND DATA-REQUEST RECOMMENDATIONS\n# ============================================================================\n\ntransform T_generate_response_knowledge :\n  intent: generate-knowledge\n  manner: \"formulate-statements and recommendations\"\n  input: P_observed_diffusion, P_mechanism_evidence,\n         K_hypothesis_assessment, K_relevance_quality,\n         M_mechanism_models, D_zones, D_geography\n  output: K_outbreak_assessment, K_response_recommendations,\n          K_confirmation_data_needs\n  actor: human\n  description: \"Formulate a public-health assessment that states candidate\n    origin zones, ranked transmission explanations, confidence and caveats,\n    recommended deployment actions, and the additional evidence needed to\n    validate the conclusions\"\n\nartifact K_outbreak_assessment : knowledge(P_observed_diffusion,\n    P_mechanism_evidence, M_mechanism_models)\n  representation form: \"qualified situational assessment\"\n  description: \"Candidate outbreak origin area or areas, timing of emergence,\n    ranked transmission explanations, supporting evidence, contradictory\n    evidence, and explicit uncertainty\"\n\nartifact K_response_recommendations : knowledge(K_outbreak_assessment,\n    D_zones, D_geography)\n  representation form: \"prioritised emergency-response recommendations\"\n  description: \"Recommended deployment of testing, clinical capacity,\n    communications, environmental investigation, and surveillance resources\n    conditional on the best-supported mechanism and affected zones\"\n\nartifact K_confirmation_data_needs : knowledge(K_outbreak_assessment)\n  representation form: \"data-collection recommendations\"\n  description: \"Required confirmation data, such as confirmed case and death\n    line lists, laboratory results, contact information, mobility data,\n    water-quality samples, water-network information, and environmental\n    sampling results\"\n```",
      "role": "assistant"
    }
  ]
}