Welcome to the InfoSci Platform
Reference Hub3
Recognition of Chemical Entities using Pattern Matching and Functional Group Classification

Recognition of Chemical Entities using Pattern Matching and Functional Group Classification

R. Hema, T. V. Geetha
Copyright: © 2016 |Volume: 12 |Issue: 4 |Pages: 24
ISSN: 1548-3657|EISSN: 1548-3665|EISBN13: 9781466689435|DOI: 10.4018/IJIIT.2016100102
Cite Article Cite Article

MLA

Hema, R., and T. V. Geetha. "Recognition of Chemical Entities using Pattern Matching and Functional Group Classification." IJIIT vol.12, no.4 2016: pp.21-44. http://doi.org/10.4018/IJIIT.2016100102

APA

Hema, R. & Geetha, T. V. (2016). Recognition of Chemical Entities using Pattern Matching and Functional Group Classification. International Journal of Intelligent Information Technologies (IJIIT), 12(4), 21-44. http://doi.org/10.4018/IJIIT.2016100102

Chicago

Hema, R., and T. V. Geetha. "Recognition of Chemical Entities using Pattern Matching and Functional Group Classification," International Journal of Intelligent Information Technologies (IJIIT) 12, no.4: 21-44. http://doi.org/10.4018/IJIIT.2016100102

Export Reference

Mendeley
Favorite Full-Issue Download

Abstract

The two main challenges in chemical entity recognition are: (i) New chemical compounds are constantly being synthesized infinitely. (ii) High ambiguity in chemical representation in which a chemical entity is being described by different nomenclatures. Therefore, the identification and maintenance of chemical terminologies is a tough task. Since most of the existing text mining methods followed the term-based approaches, the problems of polysemy and synonymy came into the picture. So, a Named Entity Recognition (NER) system based on pattern matching in chemical domain is developed to extract the chemical entities from chemical documents. The Tf-idf and PMI association measures are used to filter out the non-chemical terms. The F-score of 92.19% is achieved for chemical NER. This proposed method is compared with the baseline method and other existing approaches. As the final step, the filtered chemical entities are classified into sixteen functional groups. The classification is done using SVM One against All multiclass classification approach and achieved the accuracy of 87%. One-way ANOVA is used to test the quality of pattern matching method with the other existing chemical NER methods.

Request Access

You do not own this content. Please login to recommend this title to your institution's librarian or purchase it from the IGI Global bookstore.