Learning to extract information from semi-structured text using a discriminative context free grammar Paul Viola and Mukund Narasimhan   View Paper (PDF)   Return to Structured data Abstract . In cases where there are multiple errors, a single user correction can be propagated to correct multiple errors automatically. Using a discriminatively trained grammar, 93.71% of all tokens are labeled correctly (compared to 88.43% for a CMM) and 72.87% of records have all tokens labeled correctly (compared to 45.29% for the CMM). BIBTEX @inproceedings{1076091,   author = {Paul Viola and Mukund Narasimhan},   title = {Learning to extract information from semi-structured text using a discriminative context free grammar},   booktitle = {SIGIR '05: Proceedings of the 28th annual international ACM SIGIR conference on Research and development in information retrieval},   year = {2005},   isbn = {1-59593-034-5},   pages = {330--337},   location = {Salvador, Brazil},   doi = {http://doi.acm.org/10.1145/1076034.1076091},   publisher = {ACM Press},   address = {New York, NY, USA},    } ©2006 Association for Computing Machinery