Global media monitoring in real-time assumes handling large amount of textual data across different languages. We propose using text mining methods together with semantic processing to identify events from media and track the media reporting on them. We will demonstrate the proposed approach on operational media monitoring system that involves a large number of data streams in multiple languages. The system comprises a Cross-lingual Global Media Monitoring platform http://EventRegistry.org for (a) collecting media information from 300,000 news and social media sources, (b) performing linguistic and semantic processing in multiple languages, (c) forming events and event sequences, (d) streaming information about events in open data formats, (e) rich visualizations, and (f) search with complex queries for analysing global social dynamics. Challenges and technical solutions were discussed.