Skip to main navigation Skip to search Skip to main content

Research on Sentiment Classification Based on Hybrid BERT Model and Hypergraph Attention Network

  • Wei Dai*
  • , Dequan Zheng
  • , Feng Yu
  • , Feng Yan
  • *Corresponding author for this work
  • Harbin University of Commerce

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

Abstract

With the rapid growth of social media comments, Sentiment Analysis (SA), as a part of Natural Language Processing (NLP), faces an urgent need for efficient analysis and extraction of vast amounts of internet data. This study proposes a sentiment classification model based on a hybrid BERT model and a hypergraph attention network (HB_HAGT), aiming to accurately identify and classify textual sentiments by combining BERT's semantic representation capabilities with the local attention mechanism of the hypergraph attention network. By constructing a hypergraph model to enrich the feature representation of the data, the model demonstrates its superiority in sentiment analysis tasks through comparative experiments and ablation studies across multiple datasets.

Original languageEnglish
Title of host publicationProceedings of 17th International Conference on Machine Learning and Computing - ICMLC 2025
EditorsLin Huang, David Greenhalgh
PublisherSpringer Science and Business Media Deutschland GmbH
Pages339-349
Number of pages11
ISBN (Print)9783031948916
DOIs
StatePublished - 2025
Externally publishedYes
Event17th International Conference on Machine Learning and Computing, ICMLC 2025 - Guangzhou, China
Duration: 14 Feb 202517 Feb 2025

Publication series

NameLecture Notes in Networks and Systems
Volume1475 LNNS
ISSN (Print)2367-3370
ISSN (Electronic)2367-3389

Conference

Conference17th International Conference on Machine Learning and Computing, ICMLC 2025
Country/TerritoryChina
CityGuangzhou
Period14/02/2517/02/25

Keywords

  • Hybrid BERT Model
  • Hypergraph Attention Network
  • Sentiment Analysis

Fingerprint

Dive into the research topics of 'Research on Sentiment Classification Based on Hybrid BERT Model and Hypergraph Attention Network'. Together they form a unique fingerprint.

Cite this