Paper: Named Entity Recognition In Biomedical Texts Using An HMM Model

ACL ID W04-1216
Title Named Entity Recognition In Biomedical Texts Using An HMM Model
Venue International Joint Workshop On Natural Language Processing In Biomedicine And Its Applications NLPBA BioNLP
Session
Year 2004
Authors

Although there exists a huge number of biomedical texts online, there is a lack of tools good enough to help people get information or knowledge from them. Named entity Recognition (NER) becomes very important for further processing like information retrieval, information extraction and knowledge discovery. We introduce a Hidden Markov Model (HMM) for NER, with a word similarity-based smoothing. Our experiment shows that the word similarity-based smoothing can improve the performance by using huge unlabeled data. While many systems have laboriously hand-coded rules for all kinds of word features, we show that word similarity is a potential method to automatically get word formation, prefix, suffix and abbreviation information automatically from biomedical texts, as well as useful word dist...