An optimization based feature extraction and machine learning techniques for biomedical named entity recognition

Abstract

Bioinformatics is a branch of computer science that deals with newlinecollecting, storing, analyzing, and transmitting biological information. newlineBioinformatics uses computer programmes to do several tasks, such as newlineextracting information from the text and analyzing the relationships between newlinethem. Biomedical Named Entity Recognition (BioNER) identifies entities newlinesuch as drugs, genes, and chemicals from biomedical text, which help in newlineinformation extraction from the domain literature. Unstructured and structured newlinedocuments processing is involved in Named Entity Recognition (NER) to newlinerecognize definite entity classes and categorize these entities into some newlinepredefined classes. Techniques used to extract those entities play a significant newlinerole in this BNER. Supervised Machine Learning (SML) approaches are used newlinein various BioNER techniques. In these SML approaches, features play an newlineessential role in enhancing the recognition process s effectiveness. A set of newlinedistinguishing and discriminating characteristics is used to identify features newlineand indicate entity occurrence. Biocurators annotating only a limited number newlineof articles also consumes more processing time. Even though various biomedical named entity identification methods are available in the literature, the usage of a large feature set, newlinebiomedical dictionary, and rule framing make recognizing biomedical entities newlinefrom the literature text more difficult. Furthermore, as most current systems newlinefailed to recognize the entities over long-term sentences from the literature newlinehas an impact on performance. This research work proposes optimized feature extraction and two new frameworks to identify the biomedical entities from the literature text. newlineFor this reason, an Enhanced System for Biomedical Named Entities newlineRecognition (EBNER) and feature extraction is proposed for biomedical newlinenamed entity recognition using an Improved Particle Swarm Optimization newline(IPSO). Enhanced System for Biomedical Named Entities Recognition is newlineproposed to facilitate biocuration using a straightforward technique and newlineclassify the elements using proposed biocuration workflow acceleration. The newlinefollowing proposed framework combined Conditional Random Field (CRF) newlinewith Support Vector Machines (SVM) has the potential to tag and classify newlinedifferent categories of biomedical entities. newline newline