Scholarlink

<p>Opinion Mining (OM), or Sentiment Analysis (SA), is the computational examination of opinions, sentiments, or emotions directed towards entities such as products, services, topics or organisations. Opinions underpin nearly all human activities and play a pivotal role in shaping and enhancing our overall quality of life. Hence, OM emerges as a vital aspect of human existence. Numerous studies have explored OM applications, approaches, and methodologies. However, these studies are dispersed across diverse sources, impeding seamless access, evaluation, and a holistic perspective on practical applications, trends, approaches, and future research directions in OM. Recognising this gap in the literature, this study endeavours to fill the void by providing a harmonised and comprehensive review of OM through a systematic literature review, aggregating pertinent papers from reputable scholarly databases and sources. The study identifies nine distinct areas of OM applications in literature, which span Business, Natural Disaster, down to Health, Education, Security and Safety domains, shedding light on the landscape of sentiment analysis. It also explores potential applications of OM in domains such as Crime Prediction, Human Resources, and Governance, among others, which offers a pointer into the diverse future possibilities of SA. Delving into methodological approach preferences, the study reveals that the Hybrid method takes centre stage, commanding a 37.1% representation, followed by the Traditional Machine Learning (ML)-27.3%, Multimodal method- 15.4%, and Deep Learning (DL)-9.7%. This study shows that while recent studies are tilting towards advanced techniques of OM, such as the Hybrid, Multimodal, DL and so on, for better results, these methods are not without some setbacks. In conclusion, the study urges further exploration of underrepresented application areas such as Security and Safety domains, including Crime Prediction, Cyber Security, and Public Service, among others, to ensure a well-rounded development in applications of OM in the literature. This comprehensive analysis provides not only valuable insights into the current state of OM but also guides future research endeavours in this dynamic field.</p>

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

Taxonomy of Opinion Mining, Approaches and Domain Applications: Future Research Direction

  • 期刊论文

  • 发表日期：

    2025年7月16日

查看全文
  • Bridget C. Ujah-Ogbuagu,
  • Ahmed O. Ameen,
  • Fransisca N. Okwueleka,
  • Blessing Ogbuokiri

摘要

Opinion Mining (OM), or Sentiment Analysis (SA), is the computational examination of opinions, sentiments, or emotions directed towards entities such as products, services, topics or organisations. Opinions underpin nearly all human activities and play a pivotal role in shaping and enhancing our overall quality of life. Hence, OM emerges as a vital aspect of human existence. Numerous studies have explored OM applications, approaches, and methodologies. However, these studies are dispersed across diverse sources, impeding seamless access, evaluation, and a holistic perspective on practical applications, trends, approaches, and future research directions in OM. Recognising this gap in the literature, this study endeavours to fill the void by providing a harmonised and comprehensive review of OM through a systematic literature review, aggregating pertinent papers from reputable scholarly databases and sources. The study identifies nine distinct areas of OM applications in literature, which span Business, Natural Disaster, down to Health, Education, Security and Safety domains, shedding light on the landscape of sentiment analysis. It also explores potential applications of OM in domains such as Crime Prediction, Human Resources, and Governance, among others, which offers a pointer into the diverse future possibilities of SA. Delving into methodological approach preferences, the study reveals that the Hybrid method takes centre stage, commanding a 37.1% representation, followed by the Traditional Machine Learning (ML)-27.3%, Multimodal method- 15.4%, and Deep Learning (DL)-9.7%. This study shows that while recent studies are tilting towards advanced techniques of OM, such as the Hybrid, Multimodal, DL and so on, for better results, these methods are not without some setbacks. In conclusion, the study urges further exploration of underrepresented application areas such as Security and Safety domains, including Crime Prediction, Cyber Security, and Public Service, among others, to ensure a well-rounded development in applications of OM in the literature. This comprehensive analysis provides not only valuable insights into the current state of OM but also guides future research endeavours in this dynamic field.