Study On Collaborative Filtering Based On Weighted Interestingness

With the application and promotion of network, which offers us convenient delivery and service of the information. E-commerce developed and expanded quickly ,but it also encountered the overloading information phenomena.When the customers confront a great deal of information of commodities,it is too difficult to find their fancy goods swiftly.In order to solve the problem, E-commerce recommendation was borned at the right moment.E-commerce recommendation is the key to the lock,because it undertakes the task on distinguishment of customer's fancy and role of salesman who is responsible for introducing the commodity and giving effctive and pratical help to the cumstomers.Among the recommendation techniques, collaborative filtering is the most successive one so far.But it encounters the cold start and data sparse which Hinder the development of the technique. Much efforts is aimed at conquering it.This paper is also studied on the improving of the recommendation algorithm related to the knowledge of data mining. It introduces the conception of"associated rule on fancy measure"and"ontology clustering analysis"to excavate fancy of customers,in order that the creative recommendation takes cusotomers better experience on purchasing goods.The datas are tested on the base of new recommendation algorithm.It was proved to have a better performance of accuracy and integrity on recommendation system,specially on the problem of cold start and data sparse.