reconcile_facts
Cross-checks LLM-extracted facts against independently extracted predicate-argument records to identify potential fabrications, omissions, or confirmations, improving story consistency.
Instructions
[nlp] LLMが章から抽出した事実(llm_facts)と、機構が独立抽出した述語-項レコードを(主語,述語)軸で突き合わせる。 llm_facts は [{subject, predicate}, ...]。 返り値: agreement(一致=確証) / llm_only_check_grounding(本文に根拠が薄い=捏造の疑い) / mechanism_only_state_possible_omission(状態の申告漏れ・高シグナル) / mechanism_only_event_possible_omission(行為の申告漏れ・死亡等を含む)。 既知実体(KB)で対象を絞り、ゼロ照応解決済みの主語のみ照合。差分は確定でなく要確認。
Tool call template
Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.
{
"id": 1,
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"arguments": {
"chapter": "<chapter>",
"llm_facts": "<llm_facts>",
"chapter_text": "<chapter_text>"
},
"name": "reconcile_facts"
}
}Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| chapter | Yes | ||
| llm_facts | Yes | ||
| chapter_text | Yes | ||
| pov_character | No |