An investigation of extra clausal Processing methods for extraction Of information from the Bio medical literature

Abstract

The volume of biomedical literature is increasing exponentially newlinelargely due to data from high throughput techniques such as micro arrays and newlinegenome sequencing Efficient methods which hasten the process of literature newlinesurvey would be a considerable asset to researchers gathering pertinent newlineinformation Text mining is one such method that gathers and integrates newlineinformation thereby helping the process of effective knowledge acquisition newlineThe focus of this dissertation is information extraction and we newlineconcentrate primarily on two key tasks in this domain As a first step we newlinedeveloped a rule based system to recognize and classify biological entities as newlinethey are the key players in biological reactions The system exploits certain newlineconstraints of English grammar as well as surface clues to recognize and newlineclassify biological entities This system is augmented by an acronym detector newlinewith a precision of 96 08 and a recall of 87 5 The overall name newlinerecognition system achieves 91 7 precision and 85 3 recall which is the newlinestate of the art in this domain Using this precise entity recognition system as well as part of newlinespeech information we detect noun groups and verb groups This formed the newlinecore of the base system which extracts the arguments of phosphorylation newlinethe kinase substrate and the site of phosphorylation found within a clause newline newline