Techniques for Multilingual Security-Related Event Extraction from Online News

This chapter presents a number of techniques for multilingual event extraction, the main task is to accurately and efficiently detect key information about security-related events from electronic news media and summarize it in the form of database-like structures. Gathering such information over time is an important task for developing global news surveillance systems, particularly in the context of security threats and mass emergencies. In particular, this chapter describes novel techniques for dealing with specific extraction tasks, including: an event type classification method based on domain-specific inference rules, an approach to event geo-tagging based on utilisation of lexico-semantic patterns, a simple method for cross-lingual event information fusion, and techniques for scoring the relevance rank of automatically extracted facts.