Urban Data-topic Communication System With Personalized Recommendation Mechanism

As an important reference for urban planning and construction,relevant legal regulations,frontier academic research trends and economic trends,urban data plays an important role in current life.However,communication and data resource sharing with urban data as the topic is still insufficient.This thesis takes the needs of urban talents as the research background,and develops an urban data-topic communication system with personalized recommendation mechanism(called "Pie Of Urban Data"),which can improve the efficiency of urban data talent communication and help to maximize the use of urban data information.This thesis clarified the user's functional requirements and performance requirements for the system,and analyzes and sorts out the business processes related to each role of the system.For the architecture of the system the B/S mode was adopted,which can realize the logic independence of the system business logic layer,data interaction layer and presentation layer,and the system scalability and module reusability are improved.The conceptual model and logical model of database were designed.The system implementation mainly uses the programming language PHP.The database uses SQL Server.HTML,CSS and other interactive page design techniques are utilized.In the "Pie Of Urban Data" system several major business functions can be implemented such as article publication,personalized article recommendation,article purchase,and point usage.On the basis of satisfying the user's reading needs,the personalized recommendation mechanism of the article was studied and realized from the two perspectives of article content and user interest.The title of the article is processed as a word segmentation,and the result of the word segmentation is merged with the content of the tag filled out by the user.The weight of the combined content in the article is analyzed by the TF-IDF method,and the keywords of the article are extracted.The similarity between the article keywords is calculated by three different similarity measures,and the articles with larger similarity are selected as the recommended content.Using the written test cases,the system was tested from two aspects,i.e.function and performance.The test results show that the "Pie Of Urban Data" system works well,and the user interface is reasonable and friendly interaction.And the system can recommend the content of interest for users,the personalized recommendations meet the expection,and achieve the design goal.