Identification of multi perspective chemical attributes through chemical literature mining

Abstract

People across the world prefer e-publishing ever since the invention of World-Wide-Web (WWW). The technology changes in all the fields are aimed at removing complexities such as systematic approach for management, easy mapping of data onto enterprise relational database fields, retrieval and extraction of useful information etc. In the field of Chemistry, scientists and chemists prefer to search specific chemical attributes in the epublished chemical literature. Our focus is on mining the chemical information of text documents for extraction of chemical attributes. The mining attributes help the Natural Language Processing (NLP) to locate and retrieve data for tasks such as information retrieval, classifications, clustering and summarization. In our thesis, we have used chemistry specific feature sets to mine the various attributes. We carried out the annotation of chemical documents with the help of rule- based and machine learning approaches. For research process we utilized orthographic and morphological feature sets of chemical entities and for chemical entity extraction we used rule based pattern newlinematching approach. The extraction had limitation and could identify only the chemical entities with previously defined patterns. newline newline