contract-knowledge-base

A example of inference in a knowledge base with Zincbase and Graphviz.

About

• What happens when you make a knowledge base from a set of contracts?
• In this project, we build a knowledge base of popular licenses that people use with GitHub Repositories
• The representation of each licence's goals are based on GitHub's interpretation
• This project in no way represents actual advice, just an exercise in Knowledge Representation and Reasoning

What is Knowledge Representation?

• Sometimes it works to hard-code any number of if-else-then conditions, but this requires an explicit declaration of each relationship and outcome
• Instead, with knowledge represetation and reasoning, we specify a knowledge base of facts and rules, then allow the system to reason for the right answer
• In this project, we specify types of contracts and types of contract goals but allow a system to reason that any particular repository extends certain types of terms and conditions
• In addition to having greater flexibility, the knowledge base can return an audit trail of why it produces an answer

The App

• Hosted on Streamlit
• App Link to A Contract Companion

The Code

• We leverage concepts from knowledge reasoning and representation (KRR) and apply object-oriented programming to create a microtheory of contracts.

• Given the micro theory of facts and rules, we build queries and allow the inference engine to provide answers.

Misc Setup

Zincbase

• We use Zincbase to build and maintain a knowledge base and logic engine
• Reference to GitRepo: https://github.com/complexdb/zincbase

Streamlit

• We use Streamlit to deploy application code
• From terminal:

streamlit run app.py

A Snapshot of the Entire Knowledge Base