Research And Implementation Of Recommendation System Based On Bilingual Ontology Mapping Of Books

With the rapid developments of Internet technology, the Internet produces huge amounts of information every day. Under such circumstance, it is harder for users to obtain the information they want. Therefore, the operation mode that the traditional web sites apply has been unable to meet the personalized needs of users under different backgrounds. In order to solve this problem, the personalized recommendation system arises at the historic moment. In personalized recommendation systems, collaborative filtering algorithm is one of the most widely applied algorithms. The collaborative filtering algorithm makes recommendation according to user's score for a project or key word similarity between the users and projects. However, the algorithm has some problems, including low degree of structured information, lack of the semantic and insufficient utilization of information. In order to solve such problems, the paper proposes a hybrid recommendation system model based on ontology. Ontology technology is introduced into the system, which makes use of OWL language to describe the user and the project information. The introduction of ontology technology enables user and project have semantic information as well as improve the level of structured information. In the process of recommendation, through the analysis of user behavior information and considering the time context, the quality of recommendation has been improved.Based on the mentioned situation, the paper mainly focuses on how to use ontology technologies to improve the accuracy of the model coverage as well as how to depict users more accurately. Specifically, the content mainly includes the following three points:(1) Study the user interest model based on the time context. The user's behavior is converted into the extents of user's interest in products. Furthermore, the concept of time context is introduced while the classical time attenuation function is utilized to reduce the weights of the user behaviors if the user had been inactive for a relatively long time, which could describe the user's portrait more precisely.(2) Study on hybrid recommendation system model based on ontology of bilingual books. The fusion of classic collaborative filtering algorithm and the algorithm based on bilingual ontology matching of cross-lingual books will be employed. These two algorithms both have advantages and disadvantages. Collaborative filtering algorithms have cold start problem. If the amount of user rating data is not large enough, the recommendation is not accurate and the recommended products would be focused on the most popular commodities. On the other hand, the recommendation algorithm based on bilingual ontology matching of books is capable of covering a lot of goods. After fusing the two algorithms, the experiment proved that accuracy and coverage rate of hybrid recommendation algorithm are both higher than those of each single algorithm.(3) Design and implement a book recommendation system based on the hybrid algorithm. The system is established and accomplished by using Java technology in the Eclipse environment. Firstly, the demand of recommendation system is analyzed. Then the system architecture, function modules together with the database is designed in detail. Lastly the main functions of the system are described as well as the main interfaces of the system are demonstrated.