bxh/app/schemas/kg_extraction_v1.schema.json

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "kg_extraction_v1",
  "title": "Unified KG Extraction Output",
  "description": "统一知识抽取输出。每次抽取必须同时考虑 Entity、Event、Concept、Relation，并保留证据链。",
  "type": "object",
  "additionalProperties": false,
  "required": [
    "entities",
    "events",
    "concepts",
    "relations",
    "statements",
    "schema_proposals",
    "evidence_links"
  ],
  "properties": {
    "entities": {
      "type": "array",
      "items": { "$ref": "#/$defs/entity" }
    },
    "events": {
      "type": "array",
      "items": { "$ref": "#/$defs/event" }
    },
    "concepts": {
      "type": "array",
      "items": { "$ref": "#/$defs/concept" }
    },
    "relations": {
      "type": "array",
      "items": { "$ref": "#/$defs/relation" }
    },
    "statements": {
      "type": "array",
      "items": { "$ref": "#/$defs/statement" }
    },
    "schema_proposals": {
      "type": "array",
      "items": { "$ref": "#/$defs/schema_proposal" }
    },
    "evidence_links": {
      "type": "array",
      "items": { "$ref": "#/$defs/evidence_link" }
    }
  },
  "$defs": {
    "confidence": {
      "type": "number",
      "minimum": 0,
      "maximum": 1
    },
    "source_span": {
      "type": "object",
      "additionalProperties": false,
      "properties": {
        "evidence_id": { "type": ["string", "integer"] },
        "quote": { "type": "string" },
        "start_char": { "type": ["integer", "null"], "minimum": 0 },
        "end_char": { "type": ["integer", "null"], "minimum": 0 }
      }
    },
    "entity": {
      "type": "object",
      "additionalProperties": false,
      "required": ["temp_id", "name", "entity_type", "confidence"],
      "properties": {
        "temp_id": { "type": "string", "description": "本次抽取内的临时 ID，例如 ent_1" },
        "name": { "type": "string" },
        "entity_type": { "type": "string", "examples": ["Place", "Area", "Organization", "Person", "Facility"] },
        "aliases": { "type": "array", "items": { "type": "string" }, "default": [] },
        "description": { "type": "string", "default": "" },
        "attributes": { "type": "object", "additionalProperties": true, "default": {} },
        "spatial": {
          "type": "object",
          "additionalProperties": false,
          "properties": {
            "lng": { "type": ["number", "null"] },
            "lat": { "type": ["number", "null"] },
            "address": { "type": ["string", "null"] },
            "adcode": { "type": ["string", "null"] },
            "h3": { "type": "object", "additionalProperties": { "type": "string" } }
          },
          "default": {}
        },
        "source_spans": { "type": "array", "items": { "$ref": "#/$defs/source_span" }, "default": [] },
        "confidence": { "$ref": "#/$defs/confidence" }
      }
    },
    "event": {
      "type": "object",
      "additionalProperties": false,
      "required": ["temp_id", "title", "event_type", "confidence"],
      "properties": {
        "temp_id": { "type": "string", "description": "本次抽取内的临时 ID，例如 evt_1" },
        "title": { "type": "string" },
        "event_type": { "type": "string", "examples": ["OpeningEvent", "HistoricalEvent", "FestivalEvent"] },
        "time_text": { "type": "string", "default": "", "description": "事件发生时间原文，发布入图时映射为 event_date/event_time" },
        "time_norm": { "type": "string", "default": "", "description": "事件时间规范值，发布入图时映射为 event_date_norm，例如 1937 或 1960-04-30" },
        "description": { "type": "string", "default": "" },
        "participants": { "type": "array", "items": { "type": "string" }, "default": [] },
        "location_ref": { "type": ["string", "null"], "description": "可指向 entity.temp_id" },
        "source_spans": { "type": "array", "items": { "$ref": "#/$defs/source_span" }, "default": [] },
        "confidence": { "$ref": "#/$defs/confidence" }
      }
    },
    "concept": {
      "type": "object",
      "additionalProperties": false,
      "required": ["temp_id", "name", "concept_type", "confidence"],
      "properties": {
        "temp_id": { "type": "string", "description": "本次抽取内的临时 ID，例如 cpt_1" },
        "name": { "type": "string" },
        "concept_type": { "type": "string", "examples": ["Scene", "Category", "Topic", "Audience", "Experience"] },
        "description": { "type": "string", "default": "" },
        "parent_concepts": { "type": "array", "items": { "type": "string" }, "default": [] },
        "source_spans": { "type": "array", "items": { "$ref": "#/$defs/source_span" }, "default": [] },
        "confidence": { "$ref": "#/$defs/confidence" }
      }
    },
    "relation": {
      "type": "object",
      "additionalProperties": false,
      "required": ["relation_type", "source_type", "target_type", "confidence"],
      "properties": {
        "relation_type": { "type": "string", "examples": ["LOCATED_IN", "HAS_CONCEPT", "HAS_EVENT", "OPERATED_BY"] },
        "source_type": { "type": "string" },
        "target_type": { "type": "string" },
        "description": { "type": "string", "default": "" },
        "inverse_relation": { "type": ["string", "null"] },
        "confidence": { "$ref": "#/$defs/confidence" }
      }
    },
    "statement": {
      "type": "object",
      "additionalProperties": false,
      "required": ["subject_ref", "predicate", "object_ref", "object_kind", "confidence"],
      "properties": {
        "subject_ref": { "type": "string", "description": "entity/event/concept temp_id 或已知 natural_key" },
        "predicate": { "type": "string" },
        "object_ref": { "type": "string" },
        "object_kind": { "type": "string", "enum": ["entity", "event", "concept", "literal"] },
        "qualifiers": { "type": "object", "additionalProperties": true, "default": {} },
        "source_spans": { "type": "array", "items": { "$ref": "#/$defs/source_span" }, "default": [] },
        "confidence": { "$ref": "#/$defs/confidence" }
      }
    },
    "schema_proposal": {
      "type": "object",
      "additionalProperties": false,
      "required": ["proposal_type", "name", "reason", "confidence"],
      "properties": {
        "proposal_type": {
          "type": "string",
          "enum": ["entity_type", "event_type", "concept_type", "relation_type", "field"]
        },
        "name": { "type": "string" },
        "source_type": { "type": ["string", "null"] },
        "target_type": { "type": ["string", "null"] },
        "value_type": { "type": ["string", "null"], "examples": ["string", "number", "date", "boolean", "geo"] },
        "reason": { "type": "string" },
        "examples": { "type": "array", "items": { "type": "string" }, "default": [] },
        "confidence": { "$ref": "#/$defs/confidence" }
      }
    },
    "evidence_link": {
      "type": "object",
      "additionalProperties": false,
      "required": ["target_ref", "evidence_id", "support_type"],
      "properties": {
        "target_ref": { "type": "string" },
        "evidence_id": { "type": ["string", "integer"] },
        "support_type": { "type": "string", "enum": ["supports", "contradicts", "mentions"] },
        "quote": { "type": "string", "default": "" }
      }
    }
  }
}