Research On User Interests Modeling Techniques For Personalized Information Recommendation

The Internet is transition from Web2.0to Web3.0. The rapid development of the Internetprovides a wealth of information for people, but also people subject to interference from a largenumber of irrelevant information. The face of a flood of information resources, relying solely onthe manual screening of valuable information has been unrealistic. Now the network stresses thattake the customers as the center, to provide personalized service for users. User has its ownspecific browsing habits and interests, if the Web site dynamically improved informationorganization of the site based on user preferences, then the user will be easy and effective accessto the information they wanted. Fast, accurate selection of the information of interest to the userfrom a huge amount of information, to provide users with personalized recommendation is anurgent need, but the traditional information filtering is difficult to meet the requirements.Scholars have begun to pay attention to user interest modeling techniques. Now user interestmodeling for personalized recommendation is an important research topic, and is also thewebsite building and online community research hotspot.User interest modeling techniques as personalized recommendation foundation and core, theaccuracy of the model of user interest determines the effectiveness of the personalizedrecommendation services. This paper studies the key issues in the user interest modeling process,mainly to complete the following tasks:(1) User interest model is based on ontology.Ontology technology as an emerging semantic technology can be a good solution to theproblem of knowledge reuse and sharing. The traditional models of user interest exist weaksemantic and they can not be reused and shared. In the paper the user model based on ontologyhave the semantic comprehension skills, understanding of user interest, and tap the potentialinterest of the user, and recommend personal information to the user.In the article domain ontology is built directly based on ODP and it is expanded by Chinese Wikipedia. ODP is the largest and the most widely used artificial directory in the Internet today. Itcan cover all the concepts in the field of user interest. This method can build domain ontologydirectly without the help of experts in the field. The obtaining of user interest:(2) A new way of modeling user interest is proposed.A new way of modeling user interest based on domain ontology is proposed. User interestmodeling process is divided into three parts, which are obtaining user interest, the representationof user interest model and the learning and updating user interest model. In part one, getting theusers interest through users browsing behavior. In part two, a new kind of user interest modelbased ontology is put forward, which is user interest concept tree. In part three, User interestmodel is updated according to user interest obtained in part one, taking the degree of user interestin the page into account. A long-term and short-term compound interest model is established. Itcan reflect user long-term interests, as well as the change of short-term interest and candistinguish short-term interest and incidental browsing behavior. The performance of the systemand the user experience can be improved.(3) A simulation experiment is made to verify the effectiveness of the user interest modelbuilt in this article.An experiment is made, which is creating a personalized information recommendationsystem based on the interest model which is proposed in the article. The accuracy andeffectiveness of the user interest model can be verified through the evaluation of recall andprecision.The user interest model built in the paper is built for text information personalizedrecommendation. The content of news portal and online social network，such as micro-blog ismostly text information. The model can make the user more efficient access to the informationthey are interested in. It can provide mobile users with a great convenience. Model based onontology due to the advantage of ontology, can share user interest mode to some extent. Themodel can greatly reduce the workload of user interest modeling in personalizedrecommendation system, and lay the foundation for more extensive personalizedrecommendation.