Ad-hoc extraction of information from documents can ensure the transparency of decisions made by an organization. Different Information Extraction methods have been applied to extract information from various domains. Most widely known methods use manually annotated training documents that require high development time. The automated training methods are not scalable to large application domains. We have developed a semi-automated knowledge-engineering method for building the knowledge-base with minimal efforts. Because our method reduces manual processing of the training data, the development process is very fast. We have developed a prototype application to extract information from the project-reports of the American Recovery and Reinvestment Act (ARRA) of 2009. The fast development process of our system, its scalability to large application domains, and its high extraction effectiveness will help the transparency of management decisions by extracting and mining relevant information.
Sheikh, Mahmudul and Conlon, Sumali
"Use of a Fast Information Extraction Method as a Decision Support Tool,"
Journal of International Technology and Information Management: Vol. 19
, Article 1.
Available at: http://scholarworks.lib.csusb.edu/jitim/vol19/iss4/1