论文信息 - Extended Feature Set for Chemical Named Entity Recognition and Indexing

Extended Feature Set for Chemical Named Entity Recognition and Indexing

The BioCreative IV CHEMDNER Task provides participants with the opportunity to compare their methods for chemical named entity recognition (NER) and indexing in a controlled environment. We contributed to this task with our previous conditional random field based system [1] extended by a number of novel general and domain-specific features. For the latter, we used features derived from two existing chemical NER systems, ChemSpot [2] and OSCAR [3], as well as various external resources. In this paper, we describe our approach and present a detailed ablation study that underlines the positive effect of domain-specific features for chemical NER.

U. Leser | T. Rocktäschel | Torsten Huber | M. Weidlich | Philippe E. Thomas

[1] Andrew McCallum,et al. Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data , 2001, ICML.

[2] Yorick Wilks,et al. A Closer Look at Skip-gram Modelling , 2006, LREC.

[3] A. Valencia,et al. Text Mining for Drugs and Chemical Compounds: Methods, Tools and Applications , 2011, Molecular informatics.

[4] Egon L. Willighagen,et al. OSCAR4: a flexible architecture for chemical text-mining , 2011, J. Cheminformatics.

[5] Ulf Leser,et al. Relation Extraction for Drug-Drug Interactions using Ensemble Learning , 2011 .

[6] Ulf Leser,et al. ChemSpot: a hybrid system for chemical named entity recognition , 2012, Bioinform..

[7] Ulf Leser,et al. WBI-NER: The impact of domain-specific features on the performance of identifying and classifying mentions of drugs , 2013, *SEMEVAL.