基于用户主题偏好的智能检索算法及实现

Algorithm and Realization of Intelligent Retrieval Based on the User Topic Preference

本文分析了用户对文献的查阅日志及用户间的关联关系,结合电力行业主题范畴表,获取用户的主题偏好。综合考虑检索相关度、用户主题偏好、文献来源权威性分析、引用关系分析等,建立新的排序模型,使结果排序更加准确,从而将与用户需求最相关的文献排到前面,提高检索功能的用户体验。基于lucene 4.3实现智能检索系统,并提供相关主题词提示、主题查询扩展、相关反馈等辅助功能。评测结果表明,该系统在检索满意度和检索效率等方面有显著提升。

The paper analyzes the correlation of documents logs and users, combines with the power industry subject category list and acquires the user＇s topic preference. Considering the retrieval relevance, userpreferences, subject analysis, authoritative literature sources and reference relationship analysis, we establish a new scheduling model to make the results more accurate, which will make the most relevant articlesof users＇ demands be on top to improve the user experience of retrieval function. It realizes the intelligentretrieval system based on lucene 4.3 and provides the auxiliary function of related subjects themes, queryexpansion and relevant feedback. The evaluation results show that the system has significant improvement in retrieval satisfaction and retrieval efficiency etc.