| 袁永旭,成韬.基于node2vec模型的多维度科研合作学者推荐研究[J].数字图书馆论坛,2024,20(12):77~86 |
| 基于node2vec模型的多维度科研合作学者推荐研究 |
| Multi-Dimensional Scholars Recommendation in Scientific Research Cooperation Based on node2vec Model |
| 投稿时间：2024-09-27 |
| DOI：10.3772/j.issn.16732286.2024.12.009 |
| 中文关键词: 科研合作；学者推荐；图嵌入；node2vec；word2vec |
| 英文关键词: Science Collaboration; Scholars Recommendation; Graph Embedding; node2vec; word2vec |
| 基金项目: |
| 作者 | 单位 | | 袁永旭 | 山西医科大学管理学院；山西医科大学图书馆 | | 成韬 | 山西医科大学管理学院 |
|
| 摘要点击次数: 53 |
| 全文下载次数: 72 |
| 中文摘要: |
| 为了更好更全面地发现研究主题契合度高且合作可能性大的学者，提出一种基于图嵌入模型的多维度学者推荐模型。首先，使用CSSCI数据库作为数据源，构建作者间全连通网络，将作者间关键词相似度、合作关系、引用关系及机构属性等维度信息转化为作者间连线的权重。其次，使用node2vec图嵌入模型对该网络进行深度学习，得到每个作者的节点坐标向量。最后，通过计算作者间向量相似度达到合作学者推荐的目的。结果表明，提出的多维图嵌入模型能够有效地完成合作学者的推荐，可以为学者提供有价值的推荐名单，从而促进学术合作和交流。 |
| 英文摘要: |
| This article proposes a multi-dimensional scholars recommendation method based on graph embedding model to recognize scholars with a high degree of relevance to the research topic and high potential for collaboration. First, the CSSCI database is used as the data source to construct a fully connected network among authors. Keyword similarity, cooperative relationship, citation relationship, and institutional attributes among authors are integrated into weights for inter author connections. Second, the node2vec graph embedding model is used to perform deep learning on the network and obtain the the node coordinate vectors of each author. Finally, comprehensive vector similarity between authors is obtained, so as to complete the recommendation. The results show that the multi-dimensional graph embedding model proposed in this paper can effectively complete the recommendation of cooperative scholars, and can provide a valuable recommendation list for scholars, so as to promote academic cooperation and communication. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
