Topic Detection: Identifying Relevant Topics in Tourism Reviews

Abstract

In the past few years, user generated content (UGC) has been taking an increasingly important role in tourism. Traveller's experiences and opinions about destinations and tourism services support potential customers in their booking decisions. Sentiments can be extracted automatically from UGC and be used as valuable input for managerial decisions. An important subtask of sentiment analysis is the task of topic detection, thus, identifying the topics or product features, like room, service, or food & drink in case of hotel reviews, the review is about. The paper presents an overall approach for extracting topics from touristic UGC, making use of different data mining techniques. The applied data mining techniques are compared and evaluated on the base of hotel reviews regarding the Swedish mountain tourism destination Åre.