Loading...
|
On this page
Work information
Title: Talent Policy Knowledge Base Category: Project Type: Policy knowledge service / Retrieval-augmented generation Focus: Talent policies, source evidence, retrieval across knowledge bases, and knowledge graphs Role: System design, customization, and deployment At a glance
Kaiyi Chen LightRAG, Python, FastAPI, React, TypeScript, PostgreSQL, pgvector 2026 Quick links
|
Talent Policy Knowledge BaseGround policy lookup in source documents and make policy relationships explorable. 
Overview
The project adapts LightRAG to talent-policy services, bringing document parsing, knowledge organization, retrieval-assisted Q&A, and administration into one system. It parses formats including PDF and DOCX, organizes policy texts, attachments, entity relationships, and source evidence, and supports separate knowledge-base configurations for different policy topics. Retrieval and evidence
The retrieval interface combines policy filters, multi-turn questions, streamed answers, and citations. Multiple retrieval modes support policy lookup, application-requirement analysis, and policy comparison. Answers connect to source citations and evidence for verification, with explicit indications when evidence is insufficient. Application requirements and support criteria still require checking the policy text and the responsible authority's requirements.
Graphs and administration
The knowledge graph presents relationships among policies, eligible applicants, application requirements, and support measures. Node filters, search, layouts, and display settings help users explore policy connections. Administration covers multiple knowledge bases, document processing, model and service configuration, and user, role, and knowledge-base permissions, supporting routine maintenance from document import to knowledge delivery. 
|