Service Recommendation-oriented Correlation Analysis Of Multi-service Personal Data

We use a lot of services in life to meet the requirements of life,work,learning and other aspects.In the process of using the services will produce a large number of personal related data,these personal data portray the preferences and habits of individuals in different ways.Although these personal data are scattered among the various services,but there is a user-centric potential association between these data,it helps to provide users with more accurate recommendation results in service recommendations.Few existing recommendation algorithms apply the user's multi-source data to the algorithm.This paper mainly studies and solves the following questions:(1)Measurement of correlation: collect the data of same active user in multiple services,extract the topic from multi-source personal data based on the LDA model,propses the method of multi-source personal data measurement based on the topic similarity,and the discory a variety of model of multi-source personal data's correlation.We foud that even personal data generated by users in different services is relevant,the model of correlation may shows different.(2)Correlation evolution analysis: form the perspective of time,this paper analyze the correlation evolution law of multi service personal data.Personal data is not static,it may change over time,the user's experience,interests,hobbies,etc.This leads the change of correlation of multi serivce data,produce different ecolution law.Measure the difference between the correlation's model of different users.Because the user's behavior is different,individual vary widely,resulting in their correlation model may exist differences,by measureing the differences between the relevant model,anylyzing the evolution laws of user correlation model over time.(3)Develep recommendation strategies: According to the number the services,develop different data fuision strategies,for the information of correlation between services and the active level of users in different services,calculates the differences in the effects of recommendations made using different data fusion strategies,get the impact of correlation on the accuracy of recommendations.Find the optimal recommendation strategy when using the same set of data for recommendation.According to the correlation between user service data and recommendation corresponding optimal strategy,find the optimal strategy for users with certain charateristics.Our study collects the data of same active user in multiple services,propses the method of multi-source personal data measurement based on the topic similarity.And then excavate a number of typical peronal data correlation model,and then anylze the evolution of evolution law of multi-source personal data's correlation model over time.The research develop six data fuision strategies,and according to these strategies,fuision different source data in service recommendation algorithm,helps develop more accurate recommendation strategies and improves th e recommendation performance.