PROTOCOL_DATASET_MANIFEST.json

{
  "dataset_name": "interaction_mechanics_segments_v1",
  "status": "GREEN",
  "source": "human-ai dialogue corpora (GPT v1, Claude v2)",
  "atomic_unit": "segment",
  "total_units": 119,
  "index_types": ["semantic", "structural"],
  "rebuild_policy": "deterministic",
  "protocol_version": "v4.23",
  "mutation_policy": "append-only",
  "notes": "Behavior-first retrieval dataset",
  "created": "2025-01-23",
  "corpora": {
    "old_gpt_v1": {
      "sessions": 6,
      "segments": 39,
      "source_path": "./corpus/old_gpt_v1/"
    },
    "old_claude_v2": {
      "sessions": 8,
      "segments": 80,
      "source_path": "./corpus/old_claude_v2/"
    }
  },
  "artifacts": {
    "segment_registry": "segment_registry.jsonl",
    "primitive_signatures": "primitive_signatures.jsonl",
    "semantic_index": "indices/semantic/faiss.index",
    "structural_index": "indices/structural/faiss.index"
  },
  "embedding_model": "intfloat/e5-small-v2",
  "embedding_dim": 384,
  "index_type": "IndexFlatIP"
}