Abstract:
This end-of-studies project focuses on the effective management of the large volume of multilingual consumer comments and reviews for companies and project leaders. Based on sentiment analysis and text mining, the study explores different approaches, such as neural networks, SVMs, logistic regression, Bayes Naive, decision trees and random forests, to process data in French and English. A detailed comparison of these methods is made to determine the most suitable for sentiment analysis and multilingual text mining. In addition, two distinct scanning methods, tf-idf and one-hot vector coding, are being tested to assess their effectiveness in analyzing multilingual data