Tamil NER - Coping with Real Time Challenges
C. S. Malarkodi, Pattabhi R. K. Rao, Sobha Lalitha Devi · 2012
This paper describes various challenges encountered while developing an automatic Named Entity Recognition (NER) using Conditional Random Fields (CRFs) for Tamil. We also discuss how we have overcome some of these challenges. Though most of the challenges in NER discussed here are common to many Indian languages, in this work the focus is on Tamil, a South Indian language belonging to Dravidian language family. The corpus used in this work is the web data. The web data consisted of news paper articles, articles on blog sites and other online web portals.