Using Text Explorer to Inform and Enhance Risk and Issue Application Development and Functionality (2019-US-30MP-222)
Aug 27, 2019 12:45 PM
| Last Modified: Oct 14, 2019 8:33 AM
Scarlett Marklin, Computer Scientist, Sandia National Laboratories Yvonne Petrova, Governance, Risk and Compliance Analyst, Sandia National Laboratories
The leadership of Sandia National Laboratories requested the development of a new “risks and issues” tool that would help users record their risk management activities with more consistency and accuracy than was possible with the existing corporate tool. To accomplish this goal, developers needed details about what each text field within the existing tool had captured from users during the previous seven years. JMP Text Explorer was used to classify and identify categories of issues, risks, actions and results using latent semantic analysis with SVD. Heavy emphasis on the use of regular expressions increased the efficiency of the data pre-processing stage, which enhanced the quality of the overall results and reduced preprocessing time. In the absence of a baseline for comparison, the accuracy of the text-derived categories was gauged line by line with input from subject matter experts. The analysis highlighted gaps and inconsistencies in the historical data and clarified misunderstandings about what should be captured by the tool, resulting in a clear path forward to configure the new risks and issues tool for higher quality and more accurate data capture.