Video Website User Interestsdetermine Model Design And Simulation Implementation

That the increasingly developed internet technology has been an inevitable part ofevery walk of our life promotes the electronic commerce (e-commerce) rapidly developwell on its way. The network advertising has become the main way of e-commerce asthe amount of online users is huge. At present most of the advertisers use advertisingstorm-which pursues amount but brings tiny advertising effect, what is worse, some ofthese storming ads are disgusting. Therefore, it is the trend that to send specificadvertisements to people who may be interested in certain products.Personalized recommendation system (PRS, Personalized Recommender Systems)in the text search field have become relatively successful application, such systemrecommend interesting resources to users by the internet preference collection, which ismuch better than to send ads blindly. The PRS applied in network advertising systemcan also get better effect. In the process of streaming video transmission, the operatorsserve as transmitter and have detail information for any video by any user. How to usesuch inforamtion is the key to extract user’s preference.PRS in text field is a more successful case, however，it is still placed in theexploring stage in video field. In this paper, some of PRS cases are first analyzed andthe PRS principles are illustrated.Then, according to the characteristics of the network video and the extractedfeatures from mass data of network video operators, we propose an ontology-based userinterest determination model, which used ontology as prototype and weighted interesttree as the template. Video category ontology, describing body, actor body as well asthe title body are constructed, combining the various features of the video field. Afterthe device has access to the video stream, we mined and calculated the correspondingvideo information with its body, then polymerization processed the interest preferenceresults, so we achieved the initial interest preferences. As the user’s interest is dynamic,so the current interest preferences should be updated their weights using the historicaluser’s interest records after achieving initial interest preferences, meanwhile the historical user’s interest preferences are also updated. The experiments show that theproposed model can perform well on the determination of the user’s interest preferences,also can recognize most types of video in the network, and possess a better generality.At last, we will elaborate on each big ontology construction rules, including theinfluence on interest preference of the principles of crowd classification, theconstruction of node keywords set and video resolution.