Henriques, Roberto André PereiraRei, Ricardo Costa DiasSousa, Mário Jorge Carvalho de2022-03-142022-03-142022-01-26http://hdl.handle.net/10362/134446In recent years Artificial Intelligence has become a core part of many businesses, from manufacturers to service providers, AI can be found helping to improve business processes as well as providing customized experiences and support for the customers. Natural language processing gives computers the ability to understand human language, recent breakthroughs in multilingual models bring us closer to overcome language barriers and achieve various tasks regardless of the language. This brings to companies the opportunity to process data and provide services to customers regardless of their language. In this dissertation, we review the progress of NLP towards multilingual text classification, our results suggest that using a machine translation to augment our corpora is a suitable approach to fine-tune multi-language models like XLM-Roberta, obtaing better results than zero-shot approaches. Our results also suggest that in domain pre-training can help to increase the performance of the classification for both monolingual and multi-language classifiersengMulti-Language Text ClassificationSupervised Machine LearningDeep LearningText ClassificationNatural Language ProcessingCustomer Supportmaster thesis202962482