The Research And Implementation Of The Dynamical Recommended Algorithm Based On User Behaviors

It’s very hard for us to choose useful information from the explosive growthinformation carried by the rapid growth Internet, which makes the RecommendedSystem born. Recommender System needs to recommend users the items theirinterested based on the date of the user’s historical behavior, and with the users interestis constantly changing.The dynamic nature of the Recommendation System is required.However, traditional recommended algorithm, mainly based on collaborative filtering,did not take full account of users interesting model, thus, cannot dynamically forecastthe changes of users interest.In this thesis, by analyzing the relationship between the user behavior and userinterest, researching the dynamic nature of the recommendation system and dynamicrecommend model based on users behaviors, a Weibo recommendation systemprototype based on user behavior was proposed. The main works of this paper are asfollows:1We Analysis the dynamic characteristic of recommended system and therelationship between user behavior and interests.The flowing major Internet userbehaviors was mainly analyzed: browsing behavior, comments behavior and ratesbehavior. By analysis user behavior, a browsing behavior interesting model based onPV was proposed and simulated with experiments. Establish user emotion and commentregression model by analysis the relationship with user emotion reflected by userscomment behavior and users interest. Propose a vague interest classification method byanalyzing the relationship between the degree of users interesting and user rate.2Established a dynamic recommendation model according to the relationship ofuser behavior and interest, and proposed the related concepts: interest pressure, interestresistance and interest flow. The concrete implementation method and calculationprocess in single-user, multi-user and dynamic multi-user of the proposed model wasanalyzed. Finally, through the experimental data from Movielens video site, analysisusers hobbies change in the process of watching a movie, and quantify the user’sinterests, then establish of user’s hobbies model and use it to recommend movies to user. Recommend N website user interested to user by analyzing user’s brows history data.The proposed model was also evaluated by the comparative analysis of relevantsimulations.3Designed a Weibo recommend system prototype based on Hadoop. Thearchitecture and main modules of the system was designed in detail, concrete functionof modules was analysis. System’s main architecture include cloud computing layer,date mining layer and cloud services layer, the cloud computing model,recommendation model and user model consist the main modules of the system. Anddesigned system class diagram, implemented the microblogging recommendation,finally, pass the corresponding test.