The Study Of Personalized Movie Recommendation Technology In Home Network

With the continuous development of VOD technology and the explosive growth of multi-media resources, personalized recommendation service is an effecti ve way to solve user's"Information Lost"on Internet.Our research is set in personalized movie recommendation system in Home-Network, including the key-technology in interactive vedio application, describtion of media data,how to get user's behavior information and how to dig user's interest, based on that, we build user's interest profile and propose the personalized recommendation algorithm.This paper mainly focuses on the following points:1.Propose and realize the framework of Personalized Movie Recommendation system in Home Network.After discussing traditional personalized service system,we propose the frame work of Home-Network-Based personalized movie recommendation system, also design and realize the modules with necessary functions,like movie model building unit ,user's interest studying unit,etc.2. In order to adapt to diverse personalized demands, realize the recommendation function based on different rulesRecently, most Internet vedio applications provide recommendation service based on only one rule,there is no other choice for user.Our system subdivides user's demands, realize user's interest model-based recommendation and moives'similarity-based recommendation respectively,has better flexibility and suitbility.3. Have an attempt in Ontology-based Movie model buildingWe simply introduct the studying situation of vedio data discrebtion,analysis the advantage and disadvantage of different methodsTinking of the movie data's characteristic and the feature of personalized application, we make use of Area ontology model to describe movie data's complex semanteme structure and concept.4. Analysis user's interactive behavior in VOD applicationWe make rearch on the user's records in the VOD application of USTC's VOD service,analysis the user's behavior information according to different vedio length and type,find the relationship between user's indirectly visiting behavior and interest degree,that will be helpful for building user profile.5. Propose Movie Ontology model-based user profile In traditional personalized system,it often take key-words vector to describe user's interest,but this model could only consider the similarity between uer's interest and recourse data without distinguishing the structure characteristics of recource data.This paper propose the user profile based on movie ontology model,that profile not only has consideration on the relationship between user's behavior and user's interest,but also analysising the relationship between different semanteme layer's concept distribution and user's preference,we also give the algorithm of user's preference degree.The simulated experiment shows this kind of user profile have better performance in movie recommending application.