摘要
互联网环境下,数字图书资源"信息过载",为基于读者认知递进性提供个性化的图书知识链条。引入WO模型构造有序的特征词集合,来改进LAD算法的主题抽取精确性,据此构建图书知识链条内的主题动态演进模型;利用AT主题模型将读者信息融入与主题的关联分析之中,以通过读者—主题动态关联矩阵的提取,确定读者关注的阅读主题范围,并基于读者—主题动态关联强的计算,精确图书推荐结果;最后,在上述动态关联计算的基础上,以界面层、个性化推荐层与数据库层等3层结构为基础,搭建图书馆个性化推荐系统的整体框架。且经过测试,结果为基于动态关联计算的推荐系统MAE最大值为0.615 6,仍低于内容推荐系统MAE最小值,推荐准确性更高,达到了预设要求。
Under the internet environment, digital book resources "information overload" provides personalized book knowledge chain based on readers′ cognitive progressivity. WO model is introduced to construct ordered feature word set to improve the accuracy of topic extraction of LAD algorithm, based on which a dynamic evolution model of topics in the knowledge chain of books is constructed, at topic model is used to integrate reader information. Based on the calculation of the strong dynamic association between readers and topics, accurate book recommendation results can be obtained. Finally, on the basis of the above-mentioned dynamic association calculation, the three-tier structure of interface layer, personalized recommendation layer and database layer is built. After testing, the results show that the maximum MAE value of the recommendation system based on dynamic association calculation is 0.615 6, which is still lower than the minimum MAE value of content recommendation system, so the recommendation accuracy is higher and meets the preset requirements.
作者
蒲晔芬
Pu Yefen(Library of Xi'an Medical College,Xi'an Medical University,Xi'an 710021,China)
出处
《电子测量技术》
2020年第18期39-42,共4页
Electronic Measurement Technology
关键词
动态关联规则
趋势度
图书馆
个性化推荐
dynamic association rules
trend degrees
library
personalized recommendation