Adapting the PULS Event Extraction Framework to Analyze Russian Text
Lidia Pivovarova, Mian Du, Roman Yangarber · Työväentutkimus Vuosikirja · 2013
This paper describes a plug-in component to extend the PULS information extraction framework to analyze Russian-language text. PULS is a comprehensive framework for information extraction (IE) that is used for analysis of news in several scenarios from English-language text and is primarily monolingual. Although monolinguality is recognized as a serious limitation, building an IE system for a new language from the bottom up is very labor-intensive. Thus, the objective of the present work is to explore whether the base framework can be extended to cover additional languages with limited effort, and to leverage the pre-existing PULS modules as far as possible, in order to accelerate the development process. The component for Russian analysis is described and its performance is evaluated on two news-analysis scenarios: epidemic surveillance and cross-border security. The approach described in the paper can be generalized to a range of heavily-inflected languages.