【摘要】 针对已有科技文献查询方法往往会遗漏"特征字符不同、但语义相似"的推荐对象的问题,从特征词语义角度研究用户兴趣偏好、科技文献特征的表示方法;提出考虑语义的科技文献与用户兴趣偏好相似性度量方法。在分析个性化科技文献查询推荐业务流程的基础上,设计个性化科技文献查询推荐原型系统ALRS,介绍ALRS的设计思路、功能模块;分析该系统核心算法CGR算法的关键数据结构;详细讨论CGR算法中"基于概念泛化的相似性度量方法"的设计思路,并给出实现该原型系统的核心代码及执行示例。更多还原

【Abstract】 Common recommendation systems often miss literatures with "different characters" but "similarity in semantic".The representation of user’s interest model and feature of science and technology literature is studied from the semantic perspective of check tags,and furthermore semantic-based similarity measure is discussed.Based on the analysis of the recommendation process,ALRS is proposed,which is the personalized recommendation prototype system of science and technology literature.This paper introduces the design idea and function module,analyzes the critical data structures of CGR algorithm,discusses the design idea of similarity measure based on concept generalization in this algorithm,and gives the core code and an execution example of algorithm.更多还原