Integration of an ILP System Progol with Relational Database
- First Step toward Efficient Integration -

One solution might be to utilize database technology. In fact, some ILP systems as well as other data mining engine have been integrated with relational database management system. In this talk, we propose an ILP system Progol which was integrated with relational database through SQL.

In particular, we show an improved induction algorithm for handling the multi-valued classification problem and non-determinate background knowledge which appears in hypothesis as constant.

Integration of an ILP System Progol with Relational Database
- First Step toward Efficient Integration -

Contents

Inductive Logic Programming & Data Mining

Scaling Up Inductive Logic Programming

Data Mining & Database Technology

Implementation of DM Algorithm in SQL

Integration of Descriptive ILP with RDB

Integration Progol with RDB -Basic Idea-

Prolog & SQL

Prolog & SQL

Prolog & SQL

Bottom Clause Construction

Bottom Clause Construction

Hypothesis Evaluation

Hypothesis Evaluation

Summary of the Naive Integration

Association Rule Mining in SQL

Association Rule Mining in SQL

Decision Tree Learning

Decision Tree Learning

Decision Tree Learning

“Group by” in DM Algorithm with DB

Problem Settings

Multiple-valued classification problem

Positive and Negative Examples

Improved Algorithm

Cover set Algorithm

Cover Set Algorithm

Cover Set Algorithm

Bottom Clause Construction

Bottom Clause Construction

Bottom Clause Construction

Bottom Clause Construction

Bottom Clause Construction

Bottom Clause Construction

Hypothesis Evaluation

Hypothesis Evaluation

Non Determinate BK with Constant

Non Determinate BK with Constant

Non Determinate BK with Constant

Non Determinate BK with Constant

Non Determinate BK with Constant

Non Determinate BK with Constant

Conclusion

Future Works