Semi-automatic financial events discovery based on lexico-semantic patterns

Abstract

Due to the market sensitivity to emerging news, investors on financial markets need to continuously monitor financial events when deciding on buying and selling equities. We propose the use of lexico-semantic patterns for financial event extraction from RSS news feeds. These patterns use financial ontologies, leveraging the commonly used lexico-syntactic patterns to a higher abstraction level, thereby enabling lexico-semantic patterns to identify more and more precisely events than lexico-syntactic patterns from text. We have developed rules based on lexico-semantic patterns used to find events, and semantic actions that allow for updating the domain ontology with the effects of the discovered events. Both the lexico-semantic patterns and the semantic actions make use of the triple paradigm that fosters their easy construction and understanding by the user. Based on precision, recall, and F1 measures, we show the effectiveness of the proposed approach.