Identifying Differences in News Coverage between Cultural/Ethnic Groups

Large-scale analysis over historical news corpora provides us with unique opportunities to examine sociological issues with respect to local and mass media. In particular, we combine the Lydia named entity recognition system with an name-based ethnicity classification engine to examine issues of ethnic and geocentric sentiment/coverage bias in newspapers. We describe new methods for ethnicity and nationality detection for news entities (people), and build on this to identify interesting temporal, geospatial, and association trends in the coverage with respect to 13 distinct cultural/ethnic/linguistic (CEL) groups.