A Survey on Relational Database Based Multi Relational Classification Algorithms

Abstract

Classification on real world database is an important task in data mining. Many classification algorithms can build model only for data in single flat file as input, whereas most of real-world data bases are stored in multiple tables and managed by relational database systems. As conversion of relational data from multiple tables into a single flat file usually causes many problems, development of multi relational classification algorithms becomes popular area of research interests. Relational database based multi relational classification algorithms aim to build a model that can predict class label of unknown tuple with the help of background table knowledge.  This method keeps database in it normalized form without distorting structure of database. This paper presents survey of existing multi relational classification algorithms based on relational database.