Association Rules Mining Based On Fuzzy Theory And Its Application On Personalized Recommendation

With the development of electronic commerce, Personalized Recommendation Services has been paid attention extensively. Personalized Recommendation Services has been considered as a typical method to deal with the problem of internet mass information. Association rule is one of the important research subject in it. There are many researchs in this area, it has a very important values not only in theory but also in applications. Now, association rules technology has been used in many areas.The algorithms research is the most important part in the research of association rules. The most famous one of them is apriori algorithms presented by Agrawal. Most algorithms are improved based on apriori algorithms. Most algorithms were presented based on items those have the same significance. But in fact different items have the different importance.The thesis integrate weighted association rule model concepts with data mining algorithms .And we compared the advantage and disadvantage between several weighted association rulesWe use the actual benefit from the practical application which attribute can bring to determine the weight of the items. Moreover we had considered fully the influence of time to the database, and has made the improvement to the weighted association rule algorithm.and has carried on the detailed explanation through the example to the algorithm.The traditional algorithm was concentrates in the accurate concept . In this paper, the fuzzy association rules algorithm to determine the membership function to discuss the issue further, and designed and implemented an improved algorithm for fuzzy association rules NFWAR(New Fuzzy Weighted Association rules).A further example analysis and the simulation test through the realistic data were done to test the algorithm performance, finally it proved the algorithm can mine the meaningful association rule, and has a good experimental result.