Identification of Chemical Entities in Patent Documents

Abstract

Abstract. Biomedical literature is an important source of information for chemical compounds. However, different representations and nomenclatures for chemical entities exist, which makes the reference of chemical entities ambiguous. Many systems already exist for gene and protein entity recognition, however very few exist for chemical entities. The main reason for this is the lack of corpus to train named entity recognition systems and perform evaluation.In this paper we present a chemical entity recognizer that uses a machine learning approach based on conditional random fields (CRF) and compare the performance with dictionary-based approaches using several terminological resources. For the training and evaluation, a gold standard of manually curated patent documents was used. While the dictionary-based systems perform well in partial identification of chemical entities, the machine learning approach performs better (10% increase in F-score in comparison to the best dictionary-based system) when identifying complete entities.

Joint model for feature selection and parameter optimization coupled with classifier ensemble in chemical mention recognition

“…For SVM experiments we use YamCha 8 toolkit along with TinySVM-0.07. 9 Here, the pairwise multi-class decision method and the polynomial kernel function are used. In YamCha, only polynomial kernel function is available.…”

“…We executed this CRF based approach with only these features on the three test data sets used in our experiments. For 2008 patent data set the system proposed in [9] attains the recall, precision and F-measure values of 47.06%, 55.01% and 50.73%, respectively. Similarly for 2009 patent dataset the same system attains the recall, precision and F-measure values of 46.94%, 54.95% and 50.63%, respectively.…”

“…We also compare our approach with the existing techniques proposed in [9,2,16]. Note that in [9] a CRF based machine learning system is developed for the chemical name identification.…”

“…The CRF based system proposed in [9] utilizes only a set of few features like stem/root word, prefixes, suffixes and digit features. We executed this CRF based approach with only these features on the three test data sets used in our experiments.…”

“…Note that in [9] a CRF based machine learning system is developed for the chemical name identification. This paper addressed only the issue of identification, and no classification was performed.…”