Research On Users' Personalized Recommendation Method Based On The Weight Of Implicit Feedback

With the arrival of Web 2.0 and the era of large data,users are facing increasing serious information overload.In order to meet rising diversified and personalized demands of users,personalized recommendations based on the explicit feedback have been widely studied and applied.However,the explicit feedback has the problems of lower richness,limited and hard ways to obtain.Therefore,the implicit feedback with acquisition of lower cost,wider application scenarios and larger data size is gradually becoming the focus of the study.For different users,implicit feedback is different in its ability to infer their preferences.Therefore,based on the existing research on implicit feedback,this paper proposes a personalized recommendation method based on the weight of implicit feedback.On the one hand,the concept named the weight of implicit feedback is presented to measure the ability that implicit feedback uses to speculate the user's preferences.The weight of the implicit feedback is calculated based on the user's behavior information and the interaction the user's behavior information and the implicit feedback.Combined the implicit feedback weight with the tag,the user's interest model is constructed to make personalized recommendation based on the implicit feedback weight.On the other hand,this paper further proposes the recommendation method based on the weight of implicit feedback and the users' attributes,which is used to alleviate the incomplete recommendation.Based on the users' attribute,the interest vector of the attribute is obtained according to the label of the relevant user,and the interest model based on the user attribute is obtained.Then a linear combination with this interest model and the interest model based on the weight of implicit feedback is obtained to make the mixed recommendation.Finally,the Last.fm dataset is used to do the validated and compared experiments.