원문정보
초록
영어
For multi-relational data mining, efficiency is always a focus of research. The main bottleneck to improve the efficiency of the algorithm is hypothesis space. This paper presents the improved multi-relational decision tree algorithm, MRDTL-2, whose efficiency is improved. First, the tuple ID propogation technology is applied to the multi-relational decision tree algorithm. Secondly, under user’s guide, when a data item is greater than the transmitting threshold, set the null relationship Ra. And transmit the primary key, the background attributes, the class label to Ra, then Ra involves in other multi-relational decision tree algorithms instead of the background relations. Finally, the paper has carried on the experiments to verify the improved multi-relational decision tree algorithm MRDTL-2.
목차
1. Introduction
2. The Thought of the Improved Algorithm
3. The Concrete Implementation Process of MRDTL-2 Algorithm
3.1. Basic Theory
3.2. The Implementation of MRDTL-2 Algorithm
4. Experiment Analysis
4.1. The Experimental Running Environment
4.2. The Experimental Data
4.3. The Experiment Process and Results
4.4. Experiment Analysis
5. Conclusion
References