Shared-Tasks

Chinese Dimensional Aspect-Based Sentiment Analysis (dimABSA)

Aspect-Based Sentiment Analysis (ABSA) is a critical NLP research topic that aims to identify the aspects of a given sentence and analyze the sentiments associated with each aspect. Compared to representing affective states as several discrete classes (i.e., polarity), the dimensional approach that represents affective states as continuous numerical values (called intensity) in multiple dimensions, such as valence-arousal (VA) space, provides more fine-grained emotional information. Therefore, we organize a Chinese dimensional ABSA shared task (dimABSA) in the SIGHAN 2024 workshop, providing fine-grained sentiment intensity prediction for each extracted aspect of a restaurant review. We have three subtasks: 1) Intensity Prediction, 2) Triplet Extraction, and 3) Quadruple Extraction. Participants will be free to choose the subtasks they wish to participate in.

More information is available online at https://dimabsa2024.github.io/

Important Dates:

• Release of training data: 1st March, 2024

• Release of test data: 20th May, 2024

• Testing results submission due: 25th May, 2024 (08:00 GMT+8)

• System description paper due: 17th June, 2024 (anywhere on Earth)

• Notification of Acceptance: 1st July, 2024.

• Camera-ready deadline: 15th July, 2024

• SIGHAN 2024 Workshop: 16th August, 2024

Shared Task Organizers: 

• Lung-Hao Lee, National Yang Ming Chiao Tung University

• Liang-Chih Yu, Yuan Ze University

• Suge Wang, Shanxi University

• Jian Liao, Shanxi University