Semantic knowledge discovery from heterogeneous data sources
EKAW'12 Proceedings of the 18th international conference on Knowledge Engineering and Knowledge Management
| Hi-index | 0.00 |
Classification is an important subject in data mining and machine learning, which has been studied extensively and has a wide range of applications. Classification based on association rules is one of the most effective classification method, whose accuracy is higher and discovered rules are easier to understand comparing with classical classification methods. However, current algorithms for classification based on association rules is single table oriented, which means they can only apply to the data stored in a single relational table. Directly applying these algorithms in multi-relational data environment will result in many problems. This paper proposes a novel algorithm MrCAR for classification based on association rules in multi-relational data environment. MrCAR mines relevant features in each table to predict the class label. Close item sets technique and Tuple ID Propagation method are used to improve the performance of the algorithm. Experimental results show that MrCAR has higher accuracy and better understandability comparing with a typical existing multi relational classification algorithm.