Research On Recommending Approach Based On Semantic Analysis For Rss Web Information Services

How to help people to get what they need from a mass of web information quickly has become a research hotpot. The appearance of RSS technology change the way in which people obtain information, which bring a new service mode to solve the problem of information overload. But, the user's individual requirements are not taken into account, RSS-based Information have some problems such as poor quality and RSS explosion. However, personalized service has formed its own theories and methods in meeting users'need. Personalized recommendation technology is the core of personalized services, which decides what should be recommended to users and also affects the quality of services. So, in this paper, we mainly focus on personalized recommendation technology, and apply it to RSS-based Information services, so as to provide high-quality personalized information services with the combination of the personalized technology and RSS technology.Most current recommending algorithms are based on Vector Space Model, which raise the problem of high dimension and sparse feature. What is more, they can't solve the problems of polysemy and synonym in text data. So, in this paper we will use a new method for recommending based on Semantic Analysis, which uses the semantic similarity to assess the similarity between the user's interest model and resource. In order to improve this idea, we have a further discussion with the data representation, semantic similarity computation and personalized recommendation algorithm. First, in order to calculating semantic similarity conveniently, we propose a new way of data representation called word list. Second, we give a method of calculating semantic similarity of word list based on the theory of computing the compound concepts'distance in Quillian's semantic net model, which can be used to compute the semantic similarity between user's profile and text or the semantic similarity between texts. Then, we propose a recommending algorithm using semantic analysis (RAUSA). This algorithm not only solves the polysemy and synonym problem, but also combines skillfully the advantages of content-based and collaboration filtering. Effectively improve the accuracy of recommending.Finally, through designing and implementing a personalized RSS-based information service platform, we apply the RAUSA to RSS-based information services, which realizes the combination of the personalized technology and RSS technology well. Our experiment on this platform validates our proposed approach.