Topic Classification for Suicidology

Computational techniques for topic classi-fication can support qualitative research by automatically applying labels in prepara-tion for qualitative analyses. This paper presents an evaluation of supervised learn-ing techniques applied to one such use case, namely that of labeling emotions, instruc-tions and information in suicide notes. We train a collection of one-versus-all binary support vector machine classifiers, using cost-sensitive learning to deal with class imbalance. The features investigated range from simple bag-of-words and n-grams over stems, to information drawn from syn-tactic dependency analysis and WordNet synonym sets. The experimental results are complemented by an analysis of systematic errors in both the output of our system and the gold-standard annotations.