Research On Personalized Recommendation Of University Library Based On Comparison Of Text Similarity

With the continuous improvement and improvement of material life,the variety of products is huge and numerous.Each online product sales platform has launched its own product recommendation system,which facilitates the use of data recorded by the purchaser on its platform,records,and history of purchases.Recommend all kinds of products that have a certain desire to buy for buyers.The university library is rich and complete,and the types and numbers of books are huge and complicated.The readers who do not have the specific target of borrowing are experiencing exponential growth in the time and effort of searching for interested books in the library.In this context,this paper studies the application of personalized library recommendation service based on text similarity comparison,analyzes the experience of popular online e-commerce platform,and opens up personalized recommendation channels for university libraries,using readers.The browsing history,collection status,borrowing history and other data automatically recommend books with the willingness to borrow for the service readers.This paper briefly introduces some relevant theoretical knowledge of data mining.On this basis,it studies and designs the personalized book recommendation system,analyzes the necessity and feasibility of personalized book recommendation system,and studies the recommendation algorithm in the university library.The application in the recommendation service,combined with the current status of the library recommendation service of the Dawn Vocational College Library,gives the recommendation system design ideas and goals based on the comparison of the book title similarity for the reader history borrowing record,and finally the experiment and result of the book recommendation system.Analyze.Fine-tuning the tools for book title segmentation,mainly reflected in the use of DOS pipeline commands for input and output redirection,adjusting some common stop words,making the word segmentation results more accurate,more suitable for the title of the book title;keyword TF-IDF weight algorithm In the application of the title of the book,the specific method is given,which makes the calculation of the weight of the title of the book more realistic and more accurate.In order to shorten the time of recommendation of the book,the independent record of the separated keyword is classified by the classification number.Index;In addition,for the reality of the batch delivery of books,the weight of each keyword is counted in the idle time,and can be directly taken out for calculation when recommended,shortening the recommended time.The final recommendation results were evaluated using the evaluation methods of the current common user satisfaction questionnaire.