Fast Relational Data Mining Query Optimization for Improving the Efficiency of Relational Data Mining Systems

Data mining is the process of building predictive or descriptive models based on a large data set, often stored in a relational database. Propositional data mining systems require that the data is converted into one single table. Relational data mining systems, on the other hand, can build models directly from the relational database. While building a model, relational data mining systems execute a huge number of queries on the database and this consumes much CPU time. In our work, we propose a number of query optimization techniques that speed up query execution. Relational data mining systems generate queries and access the data in a structured way. Our optimizations exploit this structure as much as possible. Keywords— Relational Data Mining, Machine Learning, Efficient Algorithms, Query Optimization