The advent of technology and the Internet has revolutionized consumer buying behavior, with social networking sites and websites facilitating information exchange and influencing purchase decisions through user-generated reviews. These reviews, constituting a form of "electronic word of mouth" (eWOM), often hold more sway than content produced by companies. Reviews generally trend positive, reflecting satisfaction, or negative, indicating dissatisfaction. Businesses can leverage this feedback to gauge product/service satisfaction and make improvements aligned with consumer preferences. This study aims to enhance the understanding and analysis of consumer reviews on social networks using advanced deep learning algorithms, specifically convolutional neural networks (CNN) and long short-term memory (LSTM) networks, a subset of recurrent neural networks (RNN). We evaluate the accuracy and limitations of these algorithms to refine customer review analysis methods for future applications. By analyzing customer needs, businesses can develop more optimal strategies and product development plans. Focusing on the apparel sector, particularly women's clothing, this study utilizes a dataset of 32,000 customer reviews from Kaggle to conduct sentiment analysis. The research process involves: (1) collecting data comprising ratings and reviews of clothing products, (2) applying CNN and LSTM algorithms to analyze sentiment, (3) predicting consumer trends based on analysis results, and (4) proposing improvements to enhance algorithm accuracy. The findings will provide valuable consumer insights, enabling businesses to tailor their products/services to meet customer needs, thereby increasing profitability and market share.
