Using Future Work Sentences to Explore Research Trends of Different Tasks in a Special Domain
Yuchen Qian, Zhicheng Li, Wenke Hao, Yuzhuo Wang, Chengzhi Zhang · Proceedings of the Association for Information Science and Technology · 2021
Abstract Research trend detection is an important topic for scientific researchers. Future work sentences (FWS), as direct descriptions of future research, aren't fully utilized in research trend detection. Therefore, this article uses FWS to investigate research trends of different tasks in a particular domain. Taking the conference papers in the natural language processing (NLP) field as our research objects, we obtain the FWS in each paper to build the corpus and classified them into 6 main types. After that, the task of each paper is annotated, and a task system with 29 categories is constructed to compare the FWS in different tasks. The results show that the proportion of method mentioned in FWS is the highest, and different tasks focus on different FWS types: emerging tasks need more resources, while mature tasks prefer method and application. This study provides researchers a reference to understand the research trend of specific tasks and is helpful to compare different tasks.