SCORE: Story Coherence and Retrieval Enhancement for AI Narratives
Source: https://arxiv.org/abs/2503.23512
Six versions (v1 30 Mar 2025 to v6 17 Sep 2025); the record uses the latest.
Structured fields
{
"archived": false,
"revision": {
"ref": "v6 (submitted 17 Sep 2025)",
"pinned": true,
"source_url": "https://arxiv.org/abs/2503.23512"
},
"checked_at": "2026-10-17T00:00:00.000Z",
"usefulness": "Paper proposing a framework that tracks key item statuses and generates episode summaries, then uses retrieval to find related episodes and resolve narrative inconsistencies. Directly framed around episodic continuity, but only the abstract was read and no code was linked.",
"source_type": "paper",
"availability": "available",
"source_owner": "Qiang Yi et al. (21 authors)",
"next_check_at": "2026-10-31T00:00:00.000Z",
"known_limitations": [
"Only the arXiv abstract page was read; method, datasets and baselines were not inspected.",
"No code or dataset link on the abs page.",
"Results are the authors' own."
],
"upstream_updated_at": "2025-09-17T10:22:35.000Z"
}Evidence claims
- [upstream_claim] Abstract: SCORE is a framework designed to detect and resolve narrative inconsistencies. · locator: Abstract source
- [upstream_claim] Abstract: key item statuses and episode summaries feed a RAG step that identifies related episodes. · locator: Abstract source
- [upstream_claim] Abstract: authors report improved consistency and stability versus baseline GPT models (authors' claim, not independently checked). · locator: Abstract source
- [upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.
Rights
| component | license | hosting | redistribution | modification | training | Evidence and notes |
|---|---|---|---|---|---|---|
| documentation | unknown | unknown | unknown | unknown | unknown | arXiv abs page links the arXiv non-exclusive distribution license 1.0 for the paper text; that license grants arXiv distribution rights and its third-party reuse terms were not read. |
Linked records
- Do LLM plot-hole and consistency detectors work on multi-episode series canon, not just short stories or generated text? discusses → this
- Steps, each attributed. (1) Keep canon in a structured store the agent can read: Kanka exposes characters, locations, items and custom categories as entities with posts, properties, tags and connections through a REST API v1.0 (personal access token, acts on live data, no dry-run mode); bibisco kee… cites → this
- FlawedFictions tests whether an LLM spots a plot hole in a short human-written story, using an algorithm that injects plot holes (reported: state-of-the-art LLMs struggle regardless of reasoning effort and degrade as stories get longer; LLM summaries and generated stories introduce plot holes). Con… cites → this