Sensing the Online Social Sphere Using a Sentiment Analytical Approach

Abstract

Customer online feedback in the form of user-generated content (UGC) has become one of the most contentful and influential source of information in the process of customers’ as well as suppliers’ decision making. Thus, extracting customer feedback from online platforms and detecting its sentiment as well as related topics, known as sentiment analysis or opinion mining, not surprisingly, became one of the most important and vivid research veins within the domain of web mining. This chapter first gives an overview of different approaches to tackle the problem of sentiment analysis, including simple word-list-based approaches and more complex machine learning approaches which use statistical language models or part-of-speech (POS) tagging, and discusses current applications in the field of tourism. The chapter, then describes selected sentiment analytical approaches in more detail. Sentiment detection is tackled by simple word-list-based approaches and by typical supervised learning approaches, including k-nearest neighbor, support vector machines and Naive Bayes. Additionally, topic detection is tackled by methods of unsupervised learning using cluster analysis and single value decomposition. Each of these techniques are demonstrated and validated based on a prototypical implementation as part of a destination management information system (DMIS^TM) for the leading Swedish mountain destination Åre.