Abstract | ||
---|---|---|
Bacterial named entity recognition is a challenging task in biomedical field. The task is typically modeled as a sequence labeling problem, and existing work mainly adopts discrete models such as CRF (Conditional Random Fields), requiring a large amount of hand-designed features with domain experience. To address this issue, this paper explores a neural network model for the task. We empirically study the effect of word embeddings and character embeddings on the task by extending a CRF baseline using neural networks. Results show the proposed neural network model achieves competitive performance, outperforming the current best discrete model. Meanwhile, the performance can be further improved by integrating neural and discrete features. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1109/BIBM.2018.8621206 | PROCEEDINGS 2018 IEEE INTERNATIONAL CONFERENCE ON BIOINFORMATICS AND BIOMEDICINE (BIBM) |
Keywords | Field | DocType |
bacterial, conditional random fields, neural network, discrete features, embedding | Conditional random field,Embedding,Sequence labeling,Computer science,Artificial intelligence,Artificial neural network,Named-entity recognition,Machine learning | Conference |
ISSN | Citations | PageRank |
2156-1125 | 0 | 0.34 |
References | Authors | |
0 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Yafeng Ren | 1 | 102 | 13.57 |
Hao Fei | 2 | 16 | 15.51 |
Han Ren | 3 | 19 | 11.20 |