Research Domain Configuration

domain: "guantanamo"
description: "Guantánamo Bay detention and related issues"

# Data source configuration
data_sources:
  default_path: "data/guantanamo/raw_sources/miami_herald_articles.parquet"

# Output configuration  
output:
  directory: "data/guantanamo/entities"

dedup:
  similarity_thresholds:
    default: 0.75
    people: 0.82
    organizations: 0.78
    locations: 0.80
    events: 0.76
  
  lexical_blocking:
    enabled: true
    threshold: 60        # RapidFuzz score cutoff (0-100)
    max_candidates: 50   # max entities to run cosine similarity against
  
  name_variants:
    organizations:
      equivalence_groups:
        - ["Department of Defense", "Defense Department", "DoD", "Pentagon"]
        - ["American Civil Liberties Union", "ACLU"]
    locations:
      equivalence_groups:
        - ["Guantanamo Bay", "Guantanamo", "GTMO"]
        - ["United States", "U.S.", "US"]

performance:
  concurrency:
    extract_workers: 8        # parallel articles in extraction phase
    extract_per_article: 4    # parallel entity types within article
    llm_in_flight: 16         # max concurrent cloud LLM calls
    ollama_in_flight: 2       # max concurrent Ollama calls (local mode)
  queue:
    max_buffered_articles: 32  # backpressure limit for extraction → merge

# Batching configuration (for embedding calls during merge)
batching:
  embed_batch_size: 64        # texts per embedding API call
  embed_drain_timeout_ms: 100 # reserved for future async drain behaviour

cache:
  enabled: true
  embeddings:
    lru_max_items: 4096         # in-memory LRU for embedding vectors
  extraction:
    enabled: true
    subdir: "cache/extractions"  # persistent sidecar under output dir
    version: 1                   # bump to invalidate all cached extractions
  match_check:
    enabled: true
    max_items: 8192              # per-run LRU for match-checker results
  articles:
    skip_if_unchanged: true      # skip articles whose content hash hasn't changed

merge_evidence:
  max_chars: 1500         # truncate evidence text to this length
  window_chars: 240       # characters per context window around mentions
  max_windows: 3          # max context windows to extract from article

person_types:
  detainee:
    description: "A person who is or was detained at Guantánamo Bay"
    examples: ["Mohamedou Ould Slahi", "David Hicks"]
  
  lawyer:
    description: "Attorneys, legal representatives, and legal professionals"
    examples: ["Clive Stafford Smith", "Gitanjali Gutierrez"]
  
  journalist:
    description: "Reporters, writers, and media professionals"
    examples: ["Carol Rosenberg", "Andy Worthington"]

person_tags:
  civil_rights:
    description: "People involved in civil rights advocacy"
    examples: ["ACLU lawyers", "Human rights activists"]
  
  defense:
    description: "People involved in defense or military defense roles"
    examples: ["Defense attorneys", "Military defense counsel"]

# Extract People from Guantánamo Bay Articles

You are analyzing historical newspaper articles about Guantánamo Bay detention.

## Task
Extract all people mentioned in the article text. For each person, provide:

- **name**: Full name as it appears (e.g., "Carol Rosenberg", "Lt. Col. Morris Davis")
- **type**: One of: {person_types}
- **person_tags**: List of relevant tags from: {person_tags}
- **description**: Brief description of their role or relevance (1-2 sentences)
- **alternative_names**: List of alternative names, titles, or abbreviations

## Guidelines

1. **Use specific names**: Extract "Lt. Col. Morris Davis", not "military prosecutor"
2. **Include all mentions**: Even brief references count
3. **Preserve titles**: Keep military ranks, professional titles ("Dr.", "Judge")
4. **Tag accurately**: Use tags that reflect the person's role in this article
5. **No hallucinations**: Only extract people actually mentioned in the text

## Output Format

Return a JSON array of person objects. Example:

```json
[
  {
    "name": "Carol Rosenberg",
    "type": "journalist",
    "person_tags": [],
    "description": "Miami Herald reporter covering Guantanamo Bay detention issues",
    "alternative_names": []
  },
  {
    "name": "Mohamedou Ould Slahi",
    "type": "detainee",
    "person_tags": ["civil_rights"],
    "description": "Guantanamo detainee who wrote a memoir about his detention",
    "alternative_names": ["Slahi"]
  }
]

domain: "soviet_afghan_war"
description: "Soviet-Afghan War (1979-1989) military and political history"

data_sources:
  default_path: "data/soviet_afghan_war/raw_sources/articles.parquet"

output:
  directory: "data/soviet_afghan_war/entities"

dedup:
  similarity_thresholds:
    default: 0.75
    people: 0.80        # military names vary
    organizations: 0.78
    locations: 0.82     # geographic precision important
    events: 0.74        # battles have similar descriptions
  
  name_variants:
    organizations:
      equivalence_groups:
        - ["KGB", "Committee for State Security"]
        - ["Mujahideen", "Afghan Resistance"]
    locations:
      equivalence_groups:
        - ["Afghanistan", "Islamic Republic of Afghanistan"]
        - ["Kabul", "Kabul Province"]

performance:
  concurrency:
    extract_workers: 4  # conservative for local mode
    extract_per_article: 2
    llm_in_flight: 8