Research On The Recognition Of Mobile Subscribers Behavior And Personalized Recommendation Based On Multi-sources Data

With the rapid development and the wide penetration from the consumption field to the industry field of mobile Internet technology and application,it triggered a profound and significant change in the behavior of mobile subscribers.In this context,how do traditional telecom operators make use of their own large data advantages to mine the inherent law of the subscriber behavior changes?How do traditional telecom operators fully understand the needs of their subscribers and on this basis,recognize the subscribers behavior,futher more,provide personalized recommendation to the subscribers,these are the major problems to be solved at present.In view of the above problems,scholars both here and abroad have carried out corresponding research from different aspects such as theoretical model research,algorithm optimization,empirical analysis and user personalized recommendation.However,there is still no perfect research system for mobile subscriber behavior recognition,etc.Most of the researches only based on the usage behavior of traditional mobile conmmunications,and most of the usage behavior are user basic attribute class index,usage behavior index of voice communication business and consumption class index.This situation leads to the lack of study on the new trends of subscriber behavior and indicators related to innovative business.In the personalized recommendation area,there is no enough study on the comprehensive recommendation,instead of the current study on only maily a"set menu"recommendation.Therefore,this dissert:ation combs the current situation of the mobile communication industry and the challenges faced by the basic telecom operators,and makes a comparative analysis of the problem-solving paths proposed by the existing research.By taking advantage of the large-scale data advantages of the basic telecom operators,this dissertation identifies and considers the user behavior and user behavior recognition.A comprehensive analysis conducted on customer fit behavior and user personalized recommendation.From the perspective of subscriber behavior understanding and demand discovery,this dissertation provides theoretical support and method reference for basic telecom operation enterprises to carry out targeted business innovation and product optimization,and make them improve subscriber management level,and to strengthen customer and enterprise interaction,thus to achieve income promotion.The main contents of this dissertation include the following aspects:(1)Based on the analysis of the current situation of voice and data service usage of mobile subscriber,an recognization model of voice data service usage behavior of mobile subscriber is constructed based on the improved hybrid clustering algorithm,and 260,000 4G mobile subscriber are taken as samples to carry out an empirical study through Python programming.This dissertation comprehensively analyzes the current mobile subscriber’s usage behavior mode of voice and data service,service demand characteristics and consumption characteristics.(2)Based on the study of mobile Internet application behavior of mobile subscriber,a method for discoverying the community of interest is proposed,260,000 4G users’ data are taken as samples to carry out an empirical study by Python programming.The subscriber interest in mobile Internet application community is explored.(3)Combining with the characteristics of mobile communication service and subscriber behavior,this dissertation introduces the concept of customer engagement into the field of mobile communication,and conducte the identification model of customer engagement behavior of mobile subscriber with the method of grounded theory,and expound the driving mechanism of customer engagement of basic telecom operators.(4)Combining with the research of subscriber behavior recognition,a subscriber personalized recommendation model based on subscriber behavior recognition is constructed.From two aspects of the planning and management of personalized recommendation module,the strategy suggestions for mobile subscriber personalized recommendation application is put forward,and the future de-velopment direction is predicted.This dissertation mainly carried out the following innovative researches:(1)A mobile subscriber voice data service usage behavior recognition model is constructed,and through the large-scale mobile user data empirical research,get the current mobile subscriber voice data service demand characteristics and consumption characteristics.In the context of mobile Internet,data based subscribers have already replaced the traditional traffic-based subscribers.However,in the past,the pattern of recognition index of mobile subscriber’s service usage behavior were always focused on voice service,in this dissertation,the service type,consumer psychology and consumer behavior comprehensively considered in the selection of indicators,and the subscriber behavior was analyzed from different dimensions.In the aspect of algorithm,the K-means algorithm is optimized by principal component analysis(PCA),K-means++algorithm and contour coefficient(SC)algorithm,which to a certain extent complements and improves the deficiencies of current mobile subscriber service usage behavior recognition in model construction.In the aspect of model implementation,Python language is used to program and 260,000 4G subscribers’data as empirical samples was used to realize model calculation.Compared with MATLAB,Python has the characteristics of fast computing speed,low cost,high efficiency and strong expansibility,which is more in line with the current large-scale mobile subscriber data processing requirements.In this implenmentation,8 subscriber groupes were segmented.(2)Constructed a mobile Internet application usage behavior recognition model based on community discovery.Subscriber characteristics of different types of mobile Internet applications are extracted by improved TF-IDF algorithm,and the extracted features are clustered to realize the division of interest communities,which provides a theoretical and methodological support for the basic telecom operators to analyze the characteristics of subscriber Internet use behavior.In implementation,the method and model are validated by using 260000 4G users’ mobile Internet application access frequency data as empirical data sets.By combining the results of mobile subscriber’s interest community discovery with the usage behavior pattern of mobile voice data service,the characteristics of mobile subscriber usage behavior contains both service and content is obtained.The empirical results show that in 10 different types of mobile Internet applications,only video and news applications can effectively separate users.(3)Introduced the concept of research of customer engagement into the field of mobile communication,and the driving mechanism of customer engagement of the mobile subscribers is concluded by using of the grounded theory method.Mobile communication industry is a typical service industry,with the characteristics of integration of service purchase and use.Therefore,the study of mobile subscriber’s service use behavior is equivalent to the study of mobile subscriber transaction behavior,but only the study of suscriber transaction behavior is far from enough.Customer engagemen study is aiming at the non-transactional behavior of mobile subscriber.Driving mechanism of customer engagement of the mobile subscribers obtained by grounded theory research in this dissertation extended the research scope of mobile subscriber behavior.In the theoretical aspects;this dissertation expanded the study range of the mobile subscriber behavior recognition and personalized recommendation research,throgu the new model construction and improvement to the old ones.In the empirical aspect,the feasibility of the model verified by the modeling and programming of massive mobile user behavior data,and the mining of mobile subscriber behavior characteristics is realized.The results of the research can provide theoretical explanation and decision support for telecom operators to achieve precise marketing and differentiated competition.