Scholarlink

In today’s world of technology, people express their opinions and feelings through different platforms at a rapid pace, such as online discussions, social media posts, and product reviews. Sentiment analysis, also known as ABSA is crucial for making informed decisions in highly competitive markets based on customer feedback. ABSA has been utilized across various fields, including social media, product evaluations, and customer feedback. In this paper, we utilized a sentiment classification layer within our model to accurately predict the sentiment polarity of restaurant reviews using the SemEval 2014 Dataset. Our proposed model, XLNet-LGBM, achieved the highest accuracy among all baseline models, with an accuracy of 90.6 and an F1 score of 88.7. This is approximately 2.74% and 2% higher than the average accuracy and F1 score of other baseline models such as ATN-AF, IPAN-BERT, DLCF-DCA-CDM, SA-BERT, and SA-BERT-XGBoost. Therefore, our proposed model significantly outperforms the baseline models.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Hybrid Model for Aspect-Based Sentiment Analysis Using Boosting Technique

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Amit Chauhan,
  • Aman Sharma,
  • Rajni Mohana

摘要

In today’s world of technology, people express their opinions and feelings through different platforms at a rapid pace, such as online discussions, social media posts, and product reviews. Sentiment analysis, also known as ABSA is crucial for making informed decisions in highly competitive markets based on customer feedback. ABSA has been utilized across various fields, including social media, product evaluations, and customer feedback. In this paper, we utilized a sentiment classification layer within our model to accurately predict the sentiment polarity of restaurant reviews using the SemEval 2014 Dataset. Our proposed model, XLNet-LGBM, achieved the highest accuracy among all baseline models, with an accuracy of 90.6 and an F1 score of 88.7. This is approximately 2.74% and 2% higher than the average accuracy and F1 score of other baseline models such as ATN-AF, IPAN-BERT, DLCF-DCA-CDM, SA-BERT, and SA-BERT-XGBoost. Therefore, our proposed model significantly outperforms the baseline models.