Aspect-oriented sentiment analysis

• Κύριος συγγραφέας: Φρούντζος, Χρήστος
• Άλλοι συγγραφείς: Frountzos, Christos
• Γλώσσα: English
• Έκδοση: 2020
• Θέματα: Sharing economy; Sentiment analysis; Aspect-oriented; Trust; Airbnb; Non-supervised machine learning; Διαμοιραζόμενη οικονομία; Ανάλυση συναισθήματος; Πτυχές; Εμπιστοσύνη; Μη επιβλεπόμενη μηχανική μάθηση

This dissertation examines the application of the method of Aspect-Oriented Sentiment Analysis in the field of shared economy. The importance of this study arises from the significance of the concept of trust in the area of shared economy. Since a service in sharing economy is not provided by an official body or business but rather by individuals, the only means to secure future users is the experience of the previous ones, as expressed through the reviews and comments they leave on the corresponding platform. Consequently, such an environment provides the ideal conditions for Sentiment Analysis, which can contribute to maximum safety and quality of use to the users. More specifically, this paper uses user reviews of the Airbnb property rental platform to determine their emotional attitude towards the properties where they stayed, i.e. whether they were happy or not with their stay there. Their emotional attitude is considered not only as a whole, but also in terms of a number of pre-defined aspects regarding the services provided. In order to determine this attitude, a process of non-supervised machine learning is followed, which focuses on the definition of individual aspects of sentiment assessment as well as their evaluation.