Multilingual Sentiment Analysis

Grant No. SGS-2019-018 Processing of heterogeneousdata and its specialized applicationsNatural language processing (NLP) became an essential part of the artificial in-telligence field that is used daily in industry and by millions of people. Sentimentanalysis as a part of NLP is no exception. Most of research in sentiment analysis hasbeen done primarily for English, creating a significant gap in performance betweenEnglish and other languages.In this thesis, we describe the fundamental theory behind sentiment analysis. Wesummarize tasks of sentiment analysis and tasks that are related to sentiment anal-ysis. Along with a description of common basic and initial approaches, we also coverrecent state-of-the-art techniques for sentiment analysis.We provide a detailed description of common machine learning techniques used forsentiment analysis and also very recent state-of-the-art machine learning methodsand architectures of neural networks and deep learning. Further, we describe cross-lingual approaches that allow a knowledge transfer between languages.In the end, we recap our preliminary work and future direction and the aims of thefinal doctoral thesis