A Review: Text Sentiment Analysis Methods
This paper retrieves and analyzes relevant literature on text sentiment analysis from the past decade, performing a thematic analysis to summarize and categorize the mainstream methods used in sentiment analysis, and discusses their strengths and weaknesses.
Abstract
This paper conducts a literature review based on the Web of Science Core Collection and CNKI databases, employing concepts, methods, and techniques related to text sentiment analysis to construct search queries. It retrieves and analyzes relevant literature on text sentiment analysis from the past decade, performing a thematic analysis to summarize and categorize the mainstream methods used in sentiment analysis, and discusses their strengths and weaknesses. The analysis identifies three primary approaches to sentiment analysis: dictionary-based, machine learning-based, and deep learning-based methods. Each method has its merits, drawbacks, and specific application scenarios. Additionally, within deep learning, self-attention mechanisms and pre-trained models have become key research areas. The paper provides an overview of sentiment analysis methods from a broad technical perspective without delving into specific details within various fields or introducing and comparing cutting-edge methods, thus presenting some limitations. Finally, it summarizes the requirements and application scenarios for the three models and offers corresponding recommendations.