AI Workflow
AI_RETRIEVAL_X / Information-Retrieval-Tool
A TypeScript/Vite tool that drafts Boolean search queries for academic databases such as CNKI, Web of Science, PubMed, and Ei Compendex.
Purpose
Turn natural-language research needs into structured Boolean search queries for academic databases such as CNKI, Web of Science, PubMed, and Ei Compendex.
Problem
Academic retrieval is usually a translation problem: a broad research need has to become a set of precise keyword groups, Boolean operators, synonyms, and database-specific query formats. The tool is meant to reduce the first-pass friction while keeping the search process inspectable.
Role and Contribution
I built the public TypeScript/Vite project and focused on the workflow structure: how a user describes the search need, how the model drafts candidate queries, and how those queries can be adapted for common academic databases.
Current State
This is the current anchor project because it has a public repository and a concrete workflow. It should still be read as in progress: practical examples, database-specific behavior, and revision patterns need more documentation after use.
Tools and Skills
- TypeScript
- Vite
- Multi-model API support
- Academic information retrieval
- LLM-assisted workflow design
Limitations and Next Steps
The project does not claim retrieval quality benchmarks or database coverage guarantees. Next steps are to document concrete search cases, compare generated queries against manual revisions, and make limitations visible in the repository.