Text mining for wellbeing: Selecting stories using semantic and pragmatic features

Abstract

In this article, we explore an application in an area of research called wellbeing informatics. More specifically, we consider how to build a system that could be used for searching stories that relate to the interest of the user (content relevance), and help the user in his or her developmental process by providing encouragement, useful experiences, or otherwise supportive content (emotive relevance). The first objective is covered through topic modeling applying independent component analysis and the second by using sentiment analysis. We also use style analysis to exclude stories that are inappropriate in style. We discuss linguistic theories and methodological aspects of this area, outline a hybrid methodology that can be used in selecting stories that match both the content and emotive criteria, and present the results of experiments that have been used to validate the approach. © 2012 Springer-Verlag.