Validation of Scientific Paper Recommendations Methodology Based on Viewpoints of Abstracts
-
- KOBAYASHI Keita
- The University of Electro-Communications
-
- KOYAMA Kohei
- The University of Electro-Communications
-
- NARIMATSU Hiromi
- NTT Communication Science Laboratories
-
- MINAMI Yasuhiro
- The University of Electro-Communications
Bibliographic Information
- Other Title
-
- アブストラクトの観点に基づく学術論文推薦手法の検証
Description
<p>In paper recommendations, we propose a method using abstract classification to provide not only lists of papers to be read but also the reasons for the recommendations in terms of the author's viewpoints. In conventional methods, the similarity between a paper as a search query and papers that are candidates for recommendation is judged on the overall similarity, and it is difficult to provide a reason for the recommendation. In this paper, each sentence of the abstract is classified in viewpoints such as background, methods, and results, and we analyze whether the papers in the citation relationship can be classified according to each viewpoint. Then, we propose and validate methods for a paper recommendation based on the similarity of each viewpoint. The results show the effectiveness of the method based on similarity per viewpoint and the possibility of improving performance by combining it with conventional methods.</p>
Journal
-
- Proceedings of the Annual Conference of JSAI
-
Proceedings of the Annual Conference of JSAI JSAI2023 (0), 4Xin105-4Xin105, 2023
The Japanese Society for Artificial Intelligence
- Tweet
Details 詳細情報について
-
- CRID
- 1390015333244848256
-
- ISSN
- 27587347
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
-
- Abstract License Flag
- Disallowed


