Performance Enhancement of Gene Mention Tagging by Using Deep Learning and Biomedical Named Entity Recognition

Performance enhancement of biomedical named entity recognition tagging by applying a deep learning based framework is introduced with a combination of the word as well as character embedding. The input sentences first pass through the word and character level embedding, where word embedding is used to learn syntactic and semantic information and character level embedding handles those words which are out-of-vocabulary, then it passes through Bi-directional Long Short-Term Memory (BI-LSTM), where first it train the sentences in the forward direction and then again it trains the sentences in the backward direction, and finally it passes through CRF layer where the output comes in the form of gene mention tagging. The framework is tested over BioCreative II Gene mention task corpus. The deep learning framework combines with the conditional random field, and embedding techniques achieve 89.42% of F-score, which outperforms the various state-of-the-art techniques and top-ranked achieved system on BioCreative II Gene Mention competition.