Research And Implementation Of Water Information Recommender Engine Based On Hybrid Recommender Algorithm

With the rapid growth of water information resources, the field of water information facing increasingly serious problem of information overload, in order to solve the problem of water information overload and improve the utilization of water information resource, while the user does not explicitly demand they also be able to which provides information about recommended system-related information technology applications in the field of water becomes critical. In this paper, under the premise of the recommendation algorithm key technology and information resource characteristics of water depth research and analysis, the recommended water-based hybrid recommendation engine is designed and implemented. Work of this paper mainly in the following three aspects:First, by conventional algorithm recommended by the system of in-depth comparative analysis, taking into account the large amount of historical data on water information, this paper selected user-based collaborative filtering algorithm. And the addition of an improved K-means clustering algorithm and the time weighting function on the basis of user-based collaborative filtering algorithm is used to improve timeliness and accuracy of the system.Secondly, in order to solve the cold start problem user-based collaborative filtering algorithm brings, this paper used content-based recommendation method. Specifically, this paper based on TF-IDF weighting vector space model method to extract the features of users and projects and then build the users model and the recommended projects model. Thus effectively solved the cold start problem, and experimental results verify effectively improve the coverage and the overall effect of the recommendation system.Finally, based on in-depth study of mixing the recommended method of analysis, this paper selected parallel hybrid strategy with improved collaborative filtering and content-based recommendation two approaches to design and implement a water recommended engine. Initial results were recommended two recommended methods derived weighted to give the final recommendation list. The initial results of the two recommended methods derived weighted to give the final recommendation list.