What it does KGQA_HLM is a Flask app that visualizes character relationships from Dream of the Red Chamber (Hong Lou Meng) and answers questions like “Who is Lin Daiyu’s mother?” It stores triples in Neo4j, uses LTP for Chinese NLP, and renders results in a browser.

The interesting bit The project treats a classic literature course assignment as a serious NLP pipeline: spidering character profiles, building a graph database, then wiring tokenization, POS tagging, and NER into a question-answering layer. The README even carries an Anti-996 license badge.

Key highlights

• Neo4j backend with explicit create_graph.py and query_graph.py modules
• LTP (Language Technology Platform) for Chinese word segmentation and named-entity recognition
• Pre-crawled character images and JSON, so the spider need not re-run
• Five-page web UI: welcome, search, full relation map, QA, and profile display
• Python 3.6-era stack; deployment requires JDK 8 for Neo4j and manual LTP model download

Caveats

• The demo link is struck through and points to a raw IP address, so live hosting status is unclear
• LTP model paths are hardcoded; you’ll edit ltp.py before it runs
• No test suite or CI visible; this is clearly a student project, not a product

Verdict Grab it if you’re teaching or learning KGQA on Chinese text and want a complete, small-scale reference implementation. Skip if you need a maintained, production-ready graph stack.