融合Word2vec与时间因素的馆藏学术论文推荐算法  

A New Word2Vec Algorithm of Academic Paper Recommendation

作　　者：陈长华 李小涛[1,2] 邹小筑[1,2] 叶志锋[3,2] CHEN Changhua;LI Xiaotao;ZOU Xiaozhu;YE Zhifeng

机构地区：[1]南京航空航天大学图书馆 [2]南京航空航天大学工信智库评价中心 [3]南京航空航天大学

出　　处：《图书馆论坛》2019年第5期110-117,共8页Library Tribune

基　　金：江苏省图书馆学会重点课题项目"基于人工神经网络的馆藏资源精准推荐方法研究"(项目编号:17ZD02);南京航空航天大学图书馆业务创新项目"基于文本信息分析的图书精准推荐方法研究"(项目编号:TSG201704)研究成果

摘　　要：为提高图书馆文献个性化推荐的效率和用户满意度,文章提出了一个面向科研人员的个性化学术论文推荐方法:将科研人员与学术论文映射至特征空间中,基于word2vec进行建模,并通过加权相似性度量,计算待推荐论文与研究者已发表论文间的相似度,进行个性化推荐。Recommendation is the core of personalized library service. To improve the efficiency of library’s personalized literature recommendation as well as user satisfaction,this paper proposes a new personalized method of recommending academic papers for researchers. First,both the researchers and academic papers are mapped to the feature space. Subsequently,the Word2Vec algorithm is used to calculate and define the weighted similarity metric between the to-be-recommended papers and the papers published by target researchers. In the end, a personalized recommendation would be made.

关 键 词：推荐系统 学术论文 Word2Vec 

分 类 号：G250.7[文化科学—图书馆学] G252