LSTM-CNN Deep Learning Model for French Online Product Reviews Classification

Sentiment analysis (SA) is one of the most popular areas for analyzing and discovering insights from text data from various sources, including Facebook, Twitter, and Amazon. It plays a pivotal role in helping enterprises actively improve their business strategies and better understand customers' feedback on products. In this work, the dataset has been taken from Amazon, which contains French reviews. After preprocessing and extracting the features using contextualized word embedding for French-language including ELMO, ULMFiT, and CamemBERT, we applied deep learning algorithms including CNN, LSTM, and combined CNN and LSTM to classify reviews as positive or negative. The results show that the combined model (LSTM+CNN) using CamemBERT achieved the best performance to classify the French reviews with an accuracy of 93.7%.

Automated summary

Sentiment analysis is crucial in helping enterprises improve business strategies and understand customer feedback. This study used a dataset of French reviews and applied preprocessing, feature extraction, and deep learning algorithms to classify the reviews as positive or negative. Results showed that the combined model (LSTM+CNN) using CamemBERT achieved the highest accuracy of 93.7%.