Research And Design Of A Publication Recommendation System Based On Collaborative Filtering

With the development of the Internet information age and the popularization of network technology,the Internet has brought rich and colorful life to people,but also led to the problem of "information overload".The research shows that the number of scientific publications in the field of academic research is increasing at the rate of 6%~8% every year,which leads the researchers to spend a lot of time on the publication retrieval,the This indirectly leads to the increase of the cost of scientific research.Based on above problems,a publication recommendation system is designed in this paper.The traditional user-based collaborative filtering algorithm is improved,in which when calculating the user similarity,the similarity of user clicks and searching terms are integrated in a weighted way.Furthermore,when calculating score-per-click,the attenuation processing is appied to the recorded time period from the clicking to present.As the result,the neighboring users can be recalled more accurately.With the aid of the experiments,the diversity factor and the accuracy rate are chosen to be the metrics to evaluate the performance of the proposed algorithm.To compare our algorithm with the traditional user collaborative filtering algorithm,based on the recommended results,the experiment indicates that our algorithm have better performance on the diversity factor and the accuracy rate.In addition,a distributed publication grabbing system is designed in this paper.Comparing the crawl efficiency with the single crawler,the distributed crawler proposed in this paper can collect the data about publications more efficiently in a short time period,which can provide more professional information source for the follow-up system research.Finally,the interface of the push system is displayed.The system solves the cold start problem according to the professional direction of the user registration and the statistical article hits.The system also supports the custom interest bar and mail push service.The system test results show that the publication push system designed in this paper can not only increases the user experience,but also improve the research efficiency of scientific research workers to a certain extent.Therefore,the system has a broad application prospect.