I have been using natural language processing software to model student taxation learning and taxpater behavioral aspects of their approaches and interpretations of tax avoidance/evasion.
The modelling strategies used most were different types of co-occurence networks, multidimensional scaling and self-organizings maps.