Automatic data extraction: A prerequisite for productivity measurement

• Published in: 2008 IEEE International Engineering Management Conference pp. 1 - 5
• Main Authors: Zaum, D., Olbrich, M., Barke, E.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2008
• Subjects: Business; Conferences; Data mining; Industries; Knowledge engineering; Productivity; Semiconductor device measurement
• ISBN: 9781424422883; 1424422884
• ISSN: 2159-3590; 2159-3604
• DOI: 10.1109/IEMCE.2008.4617971

Improving the productivity of any business process initially requires its measurement. Therefore, automated models for comparison, simulation and analysis of products and the appendant workflows are being developed and improved constantly. Since the results delivered by such automatically trained systems are highly dependent on both quantity and quality of the input data used, gathering a statistically significant number of datasets is a prerequisite for the successful application of productivity measurement methodologies. In this paper, we present an approach to automated data extraction developed in cooperation with industry partners. Our concepts are based on the evaluation of a large collection of logfile data generated by a state-of-the-art workflow in the semiconductor industry and on staff feedback. The approach aims at providing an easy-to-use data extraction framework that can be integrated within a current work environment. The experiences gathered in the process of implementing and using our approach result in recommendations for a future unified data format for tool logfiles.