Skip to main navigation Skip to search Skip to main content

Knowledge Enhanced Target-Aware Stance Detection on Tweets

  • Faculty of Computing, Harbin Institute of Technology
  • Peng Cheng Laboratory

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Stance detection aims to determine the stance of a text towards a given target. Different from aspect-level sentiment classification, the target may not appear in the text. While existing models have achieved great success in this task using deep neural networks, their performances still drop sharply on cases where targets are not directly mentioned in texts, even with ‘target-aware’ structures. We argue that the nonalignment between targets and potentially opinioned terms in texts causes such failure and this could be remedied with external knowledge as a bridge. To this end, we propose RelNet, which leverages multiple external knowledge bases as bridges to explicitly link potentially opinioned terms in texts to targets of interest. Experiments on the well-adopted SemEval 2016 task 6 dataset demonstrate the effectiveness of the proposed model, especially on the subset where targets do not appear in texts.

Original languageEnglish
Title of host publicationKnowledge Graph and Semantic Computing
Subtitle of host publicationKnowledge Graph Empowers New Infrastructure Construction - 6th China Conference, CCKS 2021, Proceedings
EditorsBing Qin, Zhi Jin, Haofen Wang, Jeff Pan, Yongbin Liu, Bo An
PublisherSpringer Science and Business Media Deutschland GmbH
Pages171-184
Number of pages14
ISBN (Print)9789811664700
DOIs
StatePublished - 2021
Externally publishedYes
Event6th China Conference on Knowledge Graph and Semantic Computing, CCKS 2021 - Guangzhou, China
Duration: 4 Nov 20217 Nov 2021

Publication series

NameCommunications in Computer and Information Science
Volume1466 CCIS
ISSN (Print)1865-0929
ISSN (Electronic)1865-0937

Conference

Conference6th China Conference on Knowledge Graph and Semantic Computing, CCKS 2021
Country/TerritoryChina
CityGuangzhou
Period4/11/217/11/21

Keywords

  • External knowledge
  • Stance detection
  • Target-awareness

Fingerprint

Dive into the research topics of 'Knowledge Enhanced Target-Aware Stance Detection on Tweets'. Together they form a unique fingerprint.

Cite this