Sentiment analysis on Twitter
Ανάλυση συναισθήματος στο Twitter
![Thumbnail](/xmlui/bitstream/handle/unipi/15070/Gkouskos_mpsp18008.pdf.jpg?sequence=6&isAllowed=y)
Master Thesis
Author
Γκούσκος, Ευάγγελος
Gkouskos, Evangelos
Date
2022-12View/ Open
Keywords
Sentiment analysis ; Twitter ; Machine learning ; Deep learning ; Neural networks ; Ανάλυση συναισθήματος ; Νευρωνικά δίκτυα ; Μηχανική μάθησηAbstract
The rise of Web 2.0 and the convenience it brought in conjunction with Covid-19, enabled people to collaborate with each other online more than ever. This collaboration varies, from communicating through Social Media to cross-boarding selling, or buying. The capability of someone to review or to make comments about products, goods, or even political characters changed people’s behavior and needs regarding consumption and trade more generally and put communication into a new era.
Nowadays, more and more people make critics or comments on the internet. Undeniably, those reviews attracted attention from various industries in business and the academic community.
Sentiment analysis, also known as opinion mining is the scientific field that studies and analyzes reviews, critics, opinions, or even emotions and extracts the sentiment that derives (Liu, 2015).
The goal of this thesis is to analyze data from Twitter, determine their sentiment using different mechanisms and finally compare the results. Those mechanisms include machine learning algorithms and deep learning approaches. We begin with the evolution of sentiment analysis through time and in which condition we find it today. Later, we will present the algorithms that we used for this study, in addition to the preparation and preprocessing steps we did to form the data accordingly, for our task. Lastly, we conclude this study by investigating the accuracy of its algorithm and introduce new avenues and future work in the field of sentiment analysis.