Service Recommendation Method Based On Personal Mobile Service Ecosystem

Applications in the application market are experiencing an exponential growth year by year.In the face of a large number of application markets,users are in urgent need of a good recommendation mechanism to efficiently and quickly obtain applications of their own interest.Through research and analysis of the current high application market,we found that most of the existing platforms use collaborative filtering recommendation algorithm based on user group behavior similarity,ignoring the user's personality characteristics,which may not meet the user's real needs.The local application heat continues to grow,and the unpopular application will never be discovered by the user.Based on the above status,this paper has carried out a series of research around the user's personalized recommendation.During the research process,it was found that there are often rich information in the daily use of mobile phones,and it is possible to dig into the personalized data of many users,which is the basis of the full text.As a typical service form,APP presents a similar relationship with the natural ecosystem structure in the user's use process.The two interact with each other and maintain a dynamic balance in the subtle changes.A series of studies have been carried out from the following aspects:(1)Innovatively proposed the concept of the personal mobile service ecosystem model,gave a formal definition of the model,developed an automated data collection tool to collect the required data sources for generat ing the model,and generated the generated model implemented web-based visualization.(2)Since the user's personal mobile service ecosystem is in the long-term dynamic change,this paper analyzes the changes of each element in the model from the perspective of the change amount and the original data,and uses the entropy parameter to measure the change intensity.Discover and summarize the evolution patterns of domains and services,and design and implement a pattern recognition algorithm for service evolution.(3)In order to achieve the fine-grained prediction of the user's next cycle of interest preference,the prediction algorithms of the domain and service orbit are first proposed to predict the potential change of the model in the next cycle,and then the application market data is captured and constructed.The application of the word vector file converts the user's service track into an interest track to obtain the user's interest preference in the next cycle,and finally provides the user with a series of service recommendation strategies from both aspects of enriching and purifying the ecosystem.(4)Integrating the research results,designing and developing a prototype tool for personal mobile service ecosystem integrating data collection,model calculation,evolution analysis,results display and personalized recommendation,reflecting the value of the model in real-life scenarios.