Applications of Cloud-Based Cognitive Computing (Identifying Frequent Complaints from Social Media)

Abstract: This thesis explores the possibility of using IBMs cloud based software Watson for extracting complaints from social media. This is done with two different methods, both based on Watson's text parsing functionality. Different words or parts of sentences is then extracted and compared to other sources in order to determine if it can be considered a complaint, and if so, how common it is. A small test with a few test subjects and tweets resulted in very little overlap between the problems found by the system and the test subjects. As a result, Watson in its current state is not to be considered a sufficient tool for doing this type of sentiment analysis. The fact that it was cloud based did however make the process of continuously gathering and analysing information easier.