We study the task of span-level Emotion cause analysis (SECA), which is focused on extracting the specific emotion cause span(s) for a certain emotion expressed in the given context. Compared to popular clause-level emotion cause analysis (CECA), it is a finer-grained emotion cause analysis (ECA) task. The existing SECA method heavily dependents on the effectiveness of designed features, which is labor-intensive and not generalized well. In this paper, we formalize SECA as a sequence tagging task for which several variants of neural network-based sequence tagging models to extract specific emotion cause span(s) in the given context. These models combine different types of encoding and decoding approaches. Furthermore, to make our models more ``emotionally sensitive'', we utilize the multi-head attention mechanism to enhance the representation of context. Experimental evaluations conducted on two benchmark datasets show that our proposed models create new state-of-the-art results. Our work is the first using neural sequence tagging method for span-level ECA.
Span-level Emotion Cause Analysis with Neural Sequence Tagging
Xiangju Li, Wei Gao, Shi Feng, Wang Daling, and Shafiq Joty. In Proceedings of The 30th ACM International Conference on Information and Knowledge Management (CIKM'21 (short paper)) , pages xx-xx, 2021.
Abstract BibTex Slides