基于学术论文的学者研究兴趣标签发现研究

Analysis of Scholars Research Interest Tag Discovery Based on Academic Papers

标签构建对信息检索和个性化推荐有重要的辅助作用,学者的研究兴趣标签体现了一定时期内学者和某一个领域的研究热点与发展方向。以学者为研究对象,对学者的研究兴趣标签进行发现研究,有助于学者兴趣标签自动构建与推荐,对加强学术交流合作有重要作用。本文基于学术论文信息,采用LDA与Doc2Vec两种文本表示方法,对学者和兴趣标签分别进行表示,然后计算两种方法得到的学者与研究兴趣标签的余弦相似度,最终采用集成方法对兴趣标签进行融合,得到学者的研究兴趣标签。结果证明,集成方法能够获得更好地标注效果。

Tag construction plays an important role in information retrieval and personalized recommendation.Scholars’research interest tags reflect the research hotspots and development directions of scholars and a certain field in a certain period.Taking scholars as the research object,research on the scholars’research interest tags is helpful to automatic construction and recommendation of scholars’interest tags,which plays an important role in strengthening academic exchanges and cooperation.Based on the academic paper information,this paper used LDA and Doc2Vec to represent the scholars and interest tags respectively.Then,we calculated the cosine similarity between the scholars and the research interest tags obtained by these two methods.Finally,we use the ensemble method to fuse interest tags and get the research interest tags of scholars.The results turns out that the ensemble method can achieve better tagging results.