摘要
为了解决目前用户在图书馆检索系统中无法找到自己感兴趣的内容的困境,文章以抓取的各类教育网站上的课程信息作为实验数据,将用户兴趣与基于倒排索引的Lucene算法及LDA算法模型相结合,引入UCI-Lucene算法:使用Lucene算法得出基于倒排索引的搜索结果,使用LDA主题模型算法对Lucene算法得出的搜索结果计算得到课程的兴趣分布,与此同时使用LDA主题模型算法通过对用户日志进行计算得出用户的兴趣分布,接下来将课程的兴趣分布与用户的兴趣分布做相似度计算,得到课程—用户的兴趣相似度,最后通过加权Lucene算法得出的搜索结果得分和课程—用户的兴趣相似度得分得到每个课程的综合得分,得到最后的搜索排序结果。基于上述改进算法,文章设计了一款智慧图书馆个性化检索系统。实验表明,基于用户兴趣改进模型的智慧图书馆个性化检索系统不仅能够更好满足用户的搜索需求和用户兴趣,还能够显著提升搜索结果的准确性和召回率。
To address the challenge of users not being able to find content of interest in library retrieval systems,this article utilizes course information collected from various educational websites as experimental data.It combines user interests with the Lucene algorithm based on inverted indexing and introduces the UCI-Lucene algorithm.The UCILucene algorithm derives search results based on the Lucene algorithm's inverted indexing.It then employs the LDA topic modeling algorithm to calculate the interest distribution of courses from the search results obtained by the Lucene algorithm.Simultaneously,it utilizes the LDA topic modeling algorithm to calculate the interest distribution of users based on their activity logs.Next,it computes the similarity between the interest distributions of courses and users,resulting in course-user interest similarity scores.Finally,it combines the scores from the weighted Lucene algorithm and the courseuser interest similarity scores to obtain comprehensive scores for each course,yielding the final search ranking results.Based on the improved algorithm described above,the article designs a personalized retrieval system for smart libraries.Experimental results demonstrate that the user interest-enhanced model in the smart library personalized retrieval system not only better satisfies users'search needs and interests but also significantly improves the accuracy and recall of search results.
出处
《图书馆研究与工作》
2024年第2期62-69,76,共9页
Library Science Research & Work