Entity relation extraction for indigenous medical text

Abstract

Understanding the relation between entities in a vast context is a basic yet tedious task in the field of information extraction and machine translation. Entities are the prime concept in any domain, and all the information are constructed around this prime concept. The information that is to be understood is nothing but how the entity prevails in its arena and how it is supported by other entities. The task of apprehending these vital details is the backbone behind relation extraction. In this work, we present a two stage information extraction systems from English traditional medicine research articles. In the first stage, the system recognizes entities in the context by random forest-based learning which performed relatively well with an f-score of 88%. The second stage is the extraction of relation between entities using pattern recognition.