Show HN: Surfacing gaps in neuroscience and narrative insights in Final Fantasy

Category: ai-ml

Tags: knowledge-extraction, llm, pubmed, verification, pipeline

Score: 4.7/10 (Innovation: 6, Technical: 4, Documentation: 3, Utility: 4)

Crystal is a domain-agnostic knowledge extraction pipeline that uses PubMed searches and evidence tracing to verify claims before a single LLM call packages them into plain language. It emphasizes avoiding forced comparisons and only surfacing verified gaps, demonstrated through neuroscience and Final Fantasy narrative analysis. The approach is interesting for its discipline around verified inputs and structured evidence, but remains early-stage with limited documentation.

Target audience: researchers, data scientists, AI engineers

Repository: https://dmerullo.github.io/crystal-pipeline/

View on Hacker News