Bazi
Formalizing parts of Four Pillars practice as explicit calculation and interpretation rules, with attention to exceptions and source differences.
Independent research · Taiwan
Transforming Traditional Knowledge
Into Verifiable Computational Systems
Ziwei AI Lab is building a rigorous software foundation for working with traditional Chinese metaphysics: structured sources, explicit rules, reproducible computation, and explanations that can be traced back to evidence.
01 / ABOUT THE PROJECT
A research project built around clarity and evidence.
Ziwei AI Lab is an early-stage independent software research project in Taiwan, currently developed by one independent developer. It explores how traditional Chinese metaphysical knowledge can be represented as explicit, testable, and source-traceable software.
The work spans historical text processing, knowledge extraction, source verification, rule organization, and program development. Selected computation and verification functions have been implemented; the complete product is still under development and has not been commercially launched.
02 / WHAT WE'RE BUILDING
From source text to accountable software.
Three connected research tracks form the product direction.
Zi Wei Dou Shu
Organizing chart computation and rule relationships into a structure that can be tested, compared, and explained.
Knowledge infrastructure
Processing historical texts, recording provenance, modeling dependencies, and connecting computed results to their supporting sources.
03 / TECHNICAL ARCHITECTURE
Designed to make reasoning inspectable.
These are the system capabilities under design and development, not a claim that every component is complete.
01
Structured knowledge extraction
Convert source passages into normalized concepts and candidate rules.
02
Rule dependency modeling
Represent prerequisites and relationships between rules.
03
Deterministic computation
Keep calculable steps reproducible and separate from interpretation.
04
Source-grounded reasoning
Link conclusions to the passages and rule versions that support them.
05
Conflict detection
Identify inconsistent sources, assumptions, or derived results.
06
Exception handling
Make special cases explicit rather than burying them in prose.
07
Automated validation
Test calculations and rule behavior against documented cases.
08
Traceable AI explanations
Generate readable explanations tied to verified inputs and sources.
04 / CURRENT DEVELOPMENT STATUS
Early-stage Research & Development
Currently under active research, development, and validation.
Work completed to date
Literature processing, knowledge extraction, source verification, rule organization, and software development. Selected computation and validation functions are implemented.
Work still underway
Integrating the research into a complete, reliable product; expanding rule coverage; testing edge cases; and validating explanations against source material.
05 / WHY CLAUDE?
AI support for work that is difficult to scale alone.
We plan to use the Claude API to assist research and engineering workflows. It would support human-reviewed analysis and software development; deterministic calculations and source records remain essential parts of the system.
Planned integration only. Ziwei AI Lab does not claim an Anthropic partnership or a commercially launched Claude-powered product.
- 01Large-scale literature analysis
- 02Knowledge structuring
- 03Complex rule validation
- 04AI-assisted software engineering
- 05Source-traceable natural-language explanations
06 / INDEPENDENT DEVELOPMENT
Built with personal resources.
The project is being researched and developed by one independent developer. It is pre-incorporation, self-funded, and has no external venture funding. Research, validation, and development costs are meaningful constraints on the pace and scope of the work.
The current priority is to make the underlying knowledge and rules dependable before a commercial release.
07 / CONTACT
Get in touch.
For research, technical, or project inquiries: