CaseBank is a leading provider of experience-based diagnostic solutions to the Aerospace Industry. Since its inception in 1997, the company is dedicated to providing its customers with tools and solutions to manage their troubleshooting knowledge. These tools collect, organize, and share troubleshooting experience. At the heart of CaseBank’s product offering is the family of software tools called SpotLight that uses case-based reasoning technology. Using SpotLight, technicians at an airline efficiently and accurately troubleshoot complex problems for which well-defined diagnostic procedures may be unavailable. Aerospace organizations such as Bombardier, Rockwell Collins, and General Electric are among the current users of SpotLight.
CaseBank Technologies’ computing platforms comprise Windows NT 4.0 servers and client PCs running Windows 95/98 and Windows NT. Novell Netware and Microsoft Networking are the networking software. At CaseBank, TextAnalyst was installed as a standalone application on Windows 95/98 PCs.
Using a process called case base engineering, CaseBank develops high quality case bases that contain experience-based troubleshooting knowledge. The contents of case bases are synthesized from a variety of electronic and paper knowledge sources that customers have typically collected in their legacy information systems. The process of case base engineering is effort-intensive and potentially expensive. CaseBank was looking for a text-mining tool that could be used to aid the case development process and thereby reduce the case base engineering cost. CaseBank found that TextAnalyst had the potential to address the case base engineering problem. TextAnalyst was used on a number of existing electronic repositories containing troubleshooting and fault information and the case developers were able to rapidly assess the content, quality, and value of the repositories for use in case development. CaseBank estimates that TextAnalyst is likely to save case preparation time and allow it to be more responsive to its customer’s knowledge operationalization needs.
TextAnalyst was used on repositories that contained textual descriptions of problems and repair actions on aircraft. It was used to perform content analysis and create semantic networks. The network provides a overview of the important concepts and their interrelationships. Using the semantic network and the thematic structure the user can quickly develop an understanding of the repository contents. The discovered concepts provide a convenient set of topics to the user who is unfamiliar with subject matter and assist him/her in formulating meaningful queries. Automated abstracting delivers an accurate summary of the document.
TextAnalyst is intuitive and simple to use. Much can be done without reading the manual or the on-line help. Therefore, a new user can be productive in relatively short time. It is fast and robust. Tests on repositories containing over 50,000 words did not cause it to crash.
TextAnalyst cannot be embedded in other applications due to the unavailability of APIs. TextAnalyst 1.5 did not provide a means to print its semantic network or its repository contents, but the new version 1.53 includes semantic network output to MS Excel and printing.
Since the text-mining technology is still undeveloped, CaseBank’s R&D was looking for tools that would allow it explore the technology’s potential and use it in a cost-effective manner. TextAnalyst not only demonstrated the availability of the required functionality such as extraction of semantic networks and semantic network based search but also provided this functionality at an affordable price.
TextAnalyst is used at CaseBank to identify and assess the contents of electronic repositories of troubleshooting and maintenance information. Its outputs are internally used by CaseBank to select topics and documents for case base development.
Kalyan Moy Gupta, Ph.D.
Director of Research
CaseBank Technologies Inc.
1 Kenview Blvd. Brampton,
Canada, ON L6T 5E6
www.casebank.com