Recently, Deep Learning (DL) approaches have been applied to solve the
Sentiment Classification (SC) problem, which is a core task in reviews mining
or Sentiment Analysis (SA). The performances of these approaches are affected
by different factors. This paper addresses these factors and classifies them
into three categories: data preparation based factors, feature representation
based factors and the classification techniques based factors. The paper is a
comprehensive literature-based survey that compares the performance of more
than 100 DL-based SC approaches by using 21 public datasets of reviews given by
customers within three specific application domains (products, movies and
restaurants). These 21 datasets have different characteristics
(balanced/imbalanced, size, etc.) to give a global vision for our study. The
comparison explains how the proposed factors quantitatively affect the
performance of the studied DL-based SC approaches.
In this article, the authors discuss the application of Deep Learning (DL) approaches in solving the Sentiment Classification (SC) problem, which plays a crucial role in sentiment analysis and reviews mining. They aim to analyze and classify the factors that affect the performance of DL-based SC approaches into three categories: data preparation based factors, feature representation based factors, and classification techniques based factors.
The multi-disciplinary nature of this content is evident in the combination of natural language processing (NLP), machine learning, and deep learning. Sentiment analysis requires understanding and interpreting human language, which falls under NLP. Machine learning algorithms are utilized to train models on labeled data and make predictions. Deep learning techniques have recently emerged as powerful tools in this domain, providing better performance by automatically learning features from raw data.
One of the strengths of this paper is its comprehensive survey of the literature, where more than 100 DL-based SC approaches are compared using 21 public datasets from three specific application domains: products, movies, and restaurants. These datasets have diverse characteristics, including balanced/imbalanced data and different sizes, providing a holistic view of the study. By considering various datasets, the authors ensure that the findings can be generalized and applied to different contexts.
The comparison carried out in this paper not only identifies the factors that affect the performance of DL-based SC approaches but also quantifies their impact. This quantitative analysis aids in understanding the relative importance of each factor and provides insights into how to improve the performance of these approaches. The three identified categories of factors, i.e., data preparation, feature representation, and classification techniques, represent different areas where optimizations can be made.
Future research in this field could focus on exploring hybrid approaches that combine DL with other techniques such as rule-based systems or traditional machine learning algorithms. This integration could potentially enhance the overall performance and interpretability of sentiment classification models.
In conclusion, this article provides a valuable contribution to the field of sentiment analysis by analyzing the factors that influence the performance of DL-based SC approaches. Its multi-disciplinary nature and comprehensive survey make it a useful resource for researchers and practitioners working in the domain of reviews mining and sentiment analysis.