Dynamic Fuzzy Hierarchy Relational Learning

Relational learning is a new approach which is rising in the field of data mining andmachine learning in recent years. The learning task is to find out the relationalrepresentation and its search machanism, the learning rules respectively. In this paper, weused dynamic fuzzy mathematics as the theroretical basis to deal with the hierarchicalrelationship of the data. The mainly work is as follows:(1) We proposed a dynamic fuzzy logical relational learning algorithm and a dynamicfuzzy matrix hierarchical relational learning algorithm, respectively based on dynamicfuzzy logic and dynamic fuzzy matrix theory. And we had the example validations;(2) We proposed a representation method of dynamic fuzzy tree based on dynamicfuzzy set theory. Combined with the ID3algorithm, we proposed a dynamic fuzzy treehierarchical relational learning algorithm, and its application.(3) We gave the concepts of dynamic fuzzy graph, and proposed a dynamic fuzzygraph hierarchical relational learning algorithm. Also, its application is described;(4) Combined the above algorithms, we made an application on the chemicalmolecules. Results of the experiment showed the superiority of the algorithms.In summary, on the one hand, the research of this paper enriched the content ofdynamic fuzzy relational learning; on the other hand, we combined the proposedalgorithms to the application of the chemical molecules field, which is an applicationbackground for the new approach.