Using Natural Language Processing for Aftermarket Text to Increase Accuracy and Efficiency
Derek Hollingshead, Carol Parendo, Priya Peter · 2022 Annual Reliability and Maintainability Symposium (RAMS) · 2022
In aftermarket or field data, great strides are taken to analyze the numerical information within. However, the information that is contained in the text can be equally important and oftentimes takes a back seat to numerical data. There is a need to extract this valuable text data accurately and efficiently within the aftermarket systems. There are many challenges unique to text data which warrants a different approach. For example, repositories may have changed or evolved over the years, and they may contain a variety of structures. One may have drop down choices with text, others may have unstructured long field text, and within long field text you may find several ways to state something with the exact same meaning. When there is a large amount of data, interpreting text data can be lengthy and riddled with errors. To overcome this challenge for larger aftermarket data sets, a solution is to use NLP (Natural Language Processing).