Analysing Financial Vocabulary Using a New Semantic Subgroup Discovery System Hedwig

Subgroup discovery (SD) aims at finding statistically interesting subsets of objects (typically encoded as symbolic rules) for a chosen property of interest. In this paper, we present a newly developed semantic subgroup discovery system Hedwig, which exploits the ontological background knowledge to construct subgroup describing rules as well as to effectively guide the search procedure via top-down induction. We demonstrate the effectiveness of our system with an application in a financial domain: we search for interesting vocabulary patterns that accompany credit default swap (CDS) trend reversal for the financially troubled country of Portugal over a collection of over 8 million news articles collected in year and a half period. Our experiments yielded two interesting news topics that accompany CDS trend reversals of Portugal. We also provide several directions for further work.