BNEMiner: mining biomedical literature for extraction of biological target, disease and chemical entities

摘要

The paper presents a novel application to extract biomedical entities automatically using machine learning techniques from large volumes of biomedical text. The data in large quantities are accumulating day by day and requires automatic extraction of information. Data mining is the science of extracting information from large data. Biomedical Named entity recognition (BioNER) is the task of data mining that extracts named entities from biological texts. In this paper, we focus on developing a BioNER system for extraction of biological target, disease and chemical entities from biomedical texts. We developed the system using graphical based machine learning technique the CRFs. We have applied a set of diverse features containing standard lexical, syntactic and orthographic features combined with novel and biologically inspired features, action terms and process verbs. The system was evaluated with three widely recognised datasets. The results demonstrated the portability and the potency of the system.