Research On Micro-blog Online Advertising Recommender Based On A Hybrid Recommendation Algorithm

The abundant micro-blog information provides the opportunity and challenge to allthe various trades and occupations. The enterprise can be communicated effectivelywith its users or potential users in the platform. The enterprise can more accuratepositioning its own brand through the analysis of user data. However, with the surgein the number of users and instant information published, ordinary users get lost whenthey want to obtain the information which they are interested in. At the same time, theinformation which enterprise published will submerged in the vast amounts of data.As a result, users could not obtain the information.With the expanding of user number and the platform, it helps enterprise to create agood marketing environment. Micro-blog operators can be cooperated with enterprisein network marketing for new products, and new brand.This thesis attempts to help the personalized recommendation be combined withmarketing, so that we could solve this issue from the perspective of product. Unlikeother marketing for all users at the same way, the marketing model provides thepersonalized recommendation which based on user interests preference. It helpsenterprise to increase brand influence. We could establish the user interest modelbased on the analysis of micro-blog user information. According to the interest of theuser, the company’s products are recommended to the user to achieve a sustainedmarketing effect. Through the design of the product recommendation algorithm, theold and new products of the enterprise are recommended to the user according to thedifferent user preferences. Then according to the feedback information of the user, wecould evaluate the effectiveness of the recommendation algorithm.Through the research, we obtained the following research achievements andconclusions.Based on vector space model we build a user interest model. We could use theuser item rating matrix to characterized user interest. Through Bayesian classificationtechniques to categorize ad networks and using the method of classification of roughrecommended advertising to users. Then with the further applies in the classifiedrecommendation results which based on the project coordination filtration algorithm,it carries on the further recommendation to the user. We can get more preciserecommendations. Finally, the verification algorithm is feasible according to the example.