Research And Application Of Data Mining Technology Based On Mobile Communication Data

The increasing availability of location-aware technologies(GPS,LTE networks,etc.)enables us to record location history using spatio-temporal data.Such a real-world location history to some extent means the user's interest in the place and gives us an opportunity to understand the correlation between the user and the place.Based on user behavior analysis of space-time data,this paper implements a personalized friend recommendation and location recommendation system based on geographical location information.The system uniformly models each user's historical location information(space-time data)through the Hierarchy Graph Based Similarity Measurement(HGSM)framework,which can effectively measure the similarity between users and analyze the user behavior of mobile communication data,Eventually reaching three goals.First,in this recommendation system,the geographic area in a real world visited by a particular individual is taken as an implied score of the area.The second is based on the user's historical location as a measure of user similarity,and recommend to each user a potential friend of the same POI(point of interest).The third one is to assess the degree of interest in the unaccompanied area of the user by introducing the historical trajectory of the user and other users and recommend the POI(point of interest)matching the user's interest characteristics to the user.In this paper,we study the method innovation,use HGSM as the user similarity measurement coefficient,and evaluate the performance of the system through real data.It is demonstrated that the HGSM similarity measurement method is superior to the traditional forms(such as Cosine similarity and Pearson similarity degree)Therefore,compared with the traditional recommendation system,the system will provide users with a more attractive location service and a better recommendation experience.