熊回香 孟璇 叶佳鑫

收稿日期：2020-11-04

基金项目：国家社会科学基金一般项目“融合知识图谱与深度学习的在线学术资源挖掘与推荐研究”（项目编号：19BTQ005）。

作者简介：熊回香（1966-），女，教授，博士生导师，研究方向：网络信息组织与检索。孟璇（1994-），男，硕士研究生，研究方向：网络信息组织与检索。叶佳鑫（1993-），男，博士研究生，研究方向：网络信息组织与检索。

摘 要：[目的/意义]从关键词语义类型和学术文献老化两个维度出发挖掘学术论文价值，为学者推荐符合其研究需求并在时间维度上具有较大参考意义的学术论文。[方法/过程]首先，将学术论文关键词按语义类型进行划分;随后，基于共现关系计算同类型关键词间相似度，基于关键词相似度得到论文在语义类型上的相似度;然后，借用文献老化思想，计算不同类型论文的时间价值;最后，结合论文在语义类型上的相似度及时间价值，生成论文推荐列表从而进行推荐工作。[结果/结论]实证结果表明，使用该方法推荐的论文，一方面与学者研究方向相符;另一方面在时间维度上也具有较大价值，推荐的论文质量较高。

关键词：关键词;语义类型;文献老化;学术论文推荐

DOI：10.3969/j.issn.1008-0821.2021.01.002

〔中图分类号〕G252.62 〔文献标识码〕A 〔文章编号〕1008-0821（2021）01-0013-11

Recommendation of Academic Papers Based on Keyword

Semantic Type and Literature Obsolescence

Xiong Huixiang Meng Xuan Ye Jiaxin

（School of Information Management，Central China Normal University，Wuhan 430079，China）

Abstract：[Purpose/Significance]In order to recommend academic papers for scholars the value of academic papers is explored from the two dimensions of keyword semantic type and academic literature obsolescence，and these papers meet their research needs and have greater reference significance in the time dimension are recommended for.[Method/Process]First，the keywords of academic papers were divided into four types;then，the similarity between keywords of the same type was calculated based on the co-occurrence relationship，and the similarity of the papers in semantic types was obtained based on the keyword similarity;then，calculated the time value of different types of papers using the idea of literature obsolescence;finally，combined the similarity and time value of the papers in semantic types to generate a recommendation list of papers for scholars.[Result/Conclusion]The empirical result showed that，on the one hand，the papers recommended by this method were consistent with the research direction of scholars，on the other hand，they also had great value in the time dimension and high quality.

Key words：keyword;semantic type;literature obsolescence;academic paper recommendation

科學技术的不断发展，使得各学科领域内的科研成果的数量呈指数增长，而海量科研成果的出现在丰富学者所需学术资源的同时，也造成了资源检索困难，尤其是在科学研究的过程中查询和引用相关论文具有很大难度，学者难以在短时间内精准定位最具参考价值的学术论文。……