Research Of Association Rule Mining Algorithm Based On Commodity Classification Ontology In Electronic Commerce

Currently, the commercial activities carried out through the Web and bring it to facilitate the rapid development of e-commerce transaction speed has become the key driving force. We can get the knowledge of Behavior and manner of the customer groups through the web log mining. With them it can improve service, provide services personality and expend business opportunities.Association Rule Mining (ARM) is an important part in data mining. The purpose of ARM is to find the association of the different commodity in the database. With its help retailers can scientifically arrange stock, storage and shelves design. In e-commerce the method can be used in the commodity promotion.The defects of currently ARM system are as follow. The rules of the mining only based on data. Within the knowledge of the field, the count of rules is too large to get anything interesting. To solve the problem, the article proposed ARM base on the ontology (ARMO). The advantages of the method are as follow. (1) A clearer summery of the data. (2) Less rules. (3) Generalization of multilevel, more meaningful result. (4)Multi-level ARM.The article researched on three important parts of ARMO. (1) The principle of construct the commodity ontology. And the UNSPSC classification has some advantage. (2) Prune the rules with R-interesting. (3) Using improved Apriori arithmetic to implement ARMO.The article used FoodMart2000 database to validate the arithmetic. Using Java developed the arithmetic and using Jena as ontology engine. T-tree is used as data structure to implement ARMO and R-interesting is used to prune rules. At last the validity of the ARMO is validated.