Performance Analysis and Optimisation of the Fact Extractor System

摘要

DSTO has developed an in-house application called the Fact Extractor System for performing Information Extraction. This system can be used to extract interesting information from text documents. It is a component-based system providing a suite of tools to do this task. The system has a number of deployment tools, including one called FormFiller. FormFiller is an application that enables a user to process a set of documents, one at a time interactively or in an automated batch mode, with one or more Fact Extractors and save results to an output file under user control. This report describes performance testing and optimization of the FormFiller application and Fact Extractors.