THE ARCHITECTURE OF INFORMATION EXTRACTION FOR ONTOLOGY POPULATION IN CONTRACTOR SELECTION

Abstract

The enormous amount of unstructured data presents the biggest challenge to decision makers in eliciting meaningful information to support business decision-making. This study explores the potential use of ontologies in extracting and populating the information from various combinations of unstructured and semi-structured data formats such as tabular, form-based and natural language-based text. The main objective of this study is to propose an architecture of information extraction for ontology population. Contractor selection is chosen as the domain of interest. Thus, this research focuses on the extraction of contractor profiles from tender documents in order to enrich ontological contractor profile by populating the relevant extracted information. The findings are significantly good in precision and recall, in which the performance measures have reached an accuracy of 100% precision and recall for extracting information in both tabular and form-based formats. However, the precision score of relevant information extracted in natural language text is average with a percentage of 42.86% due to the limitation of the linguistic approach for processing Malay texts.

Related Papers

• → MinIE: Minimizing Facts in Open Information Extraction(2017)148 cited
• → Automatic creation of domain templates(2006)51 cited
• → When speed has a price(2013)5 cited
• → Harnessing Open Information Extraction for Entity Classification in a French Corpus(2016)4 cited
• → THE ARCHITECTURE OF INFORMATION EXTRACTION FOR ONTOLOGY POPULATION IN CONTRACTOR SELECTION(2016)