This project addresses the challenge of understanding the classical Chinese historical text "Zizhi Tongjian" by providing parallel vernacular translations and an AI-powered interactive visualization system. It targets researchers and enthusiasts, enabling deeper comprehension and data-driven exploration via structured data, relation networks, timelines, and geographical mapping.

How It Works

A pipeline extracts entities (People, Locations, Events) and Relations using LLMs from processed Markdown text. Advanced entity resolution ensures accurate disambiguation. Extracted data forms a unified knowledge base (JSON), enriched with geocoding (Amap), and visualized via a React/D3.js frontend for interactive exploration.

Quick Start & Requirements

Requires Python (data pipeline) and Node.js (frontend). Essential prerequisites include API keys for Deepseek LLM (DEEPSEEK_API_KEY) and Amap geocoding (AMAP_KEY), configured via .env. Data generation scripts must run to produce JSON files (unified_knowledge.json, juan_year_index.json) before launching the frontend (npm install, npm run dev). Online demos: https://zztj.wawuyu.cn, https://jy0284.github.io/zizhitongjian/.

Highlighted Details

• Parallel Text: Side-by-side classical and vernacular Chinese, organized by volume.
• AI Knowledge Graph: Extracts and structures historical figures, locations, events, and relationships.
• Interactive Visualization: Dynamic timeline (clustering, zoom), force-directed person-relation network, and map view (Leaflet/OpenStreetMap).
• Entity Disambiguation: Robust methods for identifying and merging distinct references to historical entities.

Maintenance & Community

Under continuous development (last update Jan 2026). Contributions welcomed via GitHub issues for text accuracy, data structure, analysis, and visualization. [todo] markers indicate areas for input. No explicit community channels listed.

Licensing & Compatibility

The repository's license is not explicitly stated, posing a significant unknown for adoption. Commercial use compatibility is unclear.

Limitations & Caveats

Full data pipeline requires external API keys (Deepseek, Amap), potentially incurring costs. Generated runtime data files must be created before local frontend launch. Text contains [todo] markers indicating content gaps.