Scholarlink

Sentiment analysis is a part of Natural language processing (NLP) which is used to examine a user’s thoughts, views, and emotional responses about a specific topic or product. Sentiment analysis on social media platforms like Facebook and Twitter has many uses and has emerged as an effective tool for understanding user sentiment. However, the difficulties in NLP are impeding sentiment analysis’s effectiveness and reliability. Deep learning models have shown promise in recent years as a way to address the difficulties in natural language processing. So, this paper investigates the different word embedding algorithms, dimensionality reduction approaches, and classification models used in the task of sentiment analysis over social media. Initially, the text related to user opinions is collected from various social media platforms. The text was then pre-processed to improve the outcomes by converting unorganized text into a format that can be efficiently used to train machine learning models. Then, word embedding approaches are utilized to convert the pre-processed text into vector form. After that, reduce the dimension of the vectors using various dimensionality reduction approaches. Finally, classifies the reduced features for analysing the sentiment of the user. From the analysis, BERT produces 86% of BERT score, ICA has 86% of NPV and GRU has 95% accuracy value. So, BERT and ICA are the most suitable word embedding and dimensionally reduction techniques with GRU classifiers for analysing the sentiment in social media platform.

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

Sentiment Analysis: Study of Various Deep Learning Approaches with Word Embedding Algorithm and Dimensionality Reduction Techniques Over Social Media

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • L. Mahalakshmi,
  • E. Anbalagan

摘要

Sentiment analysis is a part of Natural language processing (NLP) which is used to examine a user’s thoughts, views, and emotional responses about a specific topic or product. Sentiment analysis on social media platforms like Facebook and Twitter has many uses and has emerged as an effective tool for understanding user sentiment. However, the difficulties in NLP are impeding sentiment analysis’s effectiveness and reliability. Deep learning models have shown promise in recent years as a way to address the difficulties in natural language processing. So, this paper investigates the different word embedding algorithms, dimensionality reduction approaches, and classification models used in the task of sentiment analysis over social media. Initially, the text related to user opinions is collected from various social media platforms. The text was then pre-processed to improve the outcomes by converting unorganized text into a format that can be efficiently used to train machine learning models. Then, word embedding approaches are utilized to convert the pre-processed text into vector form. After that, reduce the dimension of the vectors using various dimensionality reduction approaches. Finally, classifies the reduced features for analysing the sentiment of the user. From the analysis, BERT produces 86% of BERT score, ICA has 86% of NPV and GRU has 95% accuracy value. So, BERT and ICA are the most suitable word embedding and dimensionally reduction techniques with GRU classifiers for analysing the sentiment in social media platform.