The Design And Implementation Of A College Book Recommendation System Based On Students' Learning Behaviors

The library is the core teaching and auxiliary organ of every university,and it is an important place for teachers and students to learn and improve their learning ability.When faced with millions of books,teachers and students often have difficulty finding the right books that they really need.Searching the homepage popular TOP-N recommended books is too general for personal needs and cannot play a targeted effect.Therefore,the personalized recommendation system is the new way to the development of university libraries.The traditional recommendation system is mainly based on the collaborative filtering recommendation based on the situation of students borrowing books,or the TOP-N recommendation by mining the correlation between books,which can be aimed at ordinary users.However,for specific groups,when recommending students from the same major,different learning levels,and different grades,it is often not enough to rely on the dimension of the static students' borrowing information alone,and certain student learning behaviors are required.To solve these problems,this thesis designs a customized recommendation scheme by constructing a hierarchical student model,analyzing students' learning behavior habits,and combining students' borrowing records in the library and the frequency of visiting the library.Based on them,this thesis designs and implements a college book recommendation System.The main contents are summarized as follows:(1)We build a student hierarchical model.We collect the basic information of students of the same major over the years,analyze their academic performance,scholarships,honorary awards,scientific research achievements,etc.in each semester,preprocess and normalize the relevant data,and obtain the overall learning level of the students through the weighted summation.The decision tree realizes the classification of students: excellent,good,medium,qualified,and unqualified,and.(2)We define a collaborative filtering based on the student feature model.We collect dynamic data source student access control information and borrowing information,carry out data preprocessing and normalization,and combine with the student hierarchical model to extract student characteristics from data such as student admission frequency and borrowing history,weighted and merged,and carry out student-based feature extraction.Collaborative filtering of features are formed for a set of similar students.Corresponding books are recommended for target students through the ranking of books preferred by similar students.(3)We design recommendation solutions based on student multi-goals.For the similar sets of students after collaborative filtering,we analyze and study the multi-objective needs of students,and provide customized recommendation solutions for students of the same major,different grades,different learning ability levels,non-professional routine needs,and edge data freshmen admission.