Exploring and Verbalizing Academic Ideas by Concept Co-occurrence

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 13001–13027 July 9-14, 2023 ©2023 Association for Computational Linguistics Exploring and Verbalizing Academic Ideas by Concept Co-occurrence Yi Xu1, Shuqian Sheng1, Bo Xue1, Luoyi Fu1∗, Xinbing Wang1, Chenghu Zhou2 1Shanghai Jiao Tong University, Shanghai, China 2IGSNRR, Chinese Academy of Sciences, Beijing, China {yixu98, susisheng, sappho_x, yiluofu, xwang8}@sjtu.edu.cn Abstract Researchers usually come up with new ideas only after thoroughly comprehending vast quantities of literature. The difficulty of this procedure is exacerbated by the fact that the number of academic publications is growing exponentially. In this study, we devise a frame- work based on concept co-occurrence for aca- demic idea inspiration, which has been inte- grated into a research assistant system. From our perspective, the fusion of two concepts that co-occur in an academic paper can be regarded as an important way of the emergence of a new idea. We construct evolving concept graphs according to the co-occurrence relationship of concepts from 20 disciplines or topics. Then we design a temporal link prediction method based on masked language model to explore po- tential connections between different concepts. To verbalize the newly discovered connections, we also utilize the pretrained language model to generate a description of an idea based on a new data structure called co-occurrence citation quintuple. We evaluate our proposed system using both automatic metrics and human assess- ment. The results demonstrate that our system has broad prospects and can assist researchers in expediting the process of discovering new ideas.1 1 Introduction Academic publications have witnessed the evolu- tion and advancement of human civilization. In modern society, out-of-box and interdisciplinary scientific work can get more attention from science funders, industry, and the public (Thurner et al., 2020), where a good idea is the cornerstone of aca- demic research. However, for most researchers, it takes a lot of time to put forward new ideas. For one thing, the number of academic publications is increasing exponentially, and it is difficult for an ∗ Luoyi Fu is the corresponding author. 1The project is publicly available for research purpose https://github.com/xyjigsaw/Kiscovery. independent researcher to understand these papers thoroughly. Besides, researchers often focus on their specialized but narrow fields, which makes it a challenge to discover underlying connections be- yond their familiar areas (Lahav et al., 2022; Krenn and Zeilinger, 2020). In this work, our purpose is to unveil the profound connections between different academic concepts and ignite researchers’ explo- ration of potential academic ideas while expediting the research process. The two primary goals are idea exploration and verbalization. For the first goal, we need to understand how new ideas originate. Generally speaking, the emer- gence of a simple idea is often formed by the inter- action between two different concepts rather than from scratch. For example, the combination of con- volution and graph neural network contributes to graph convolutional network (Kipf and Welling, 2017). This understanding of idea as connection and combination inspires us to model the process of …

Content truncated to 500 words