P-MATCH: Identifying Part Name in Noisy Text Data

Abstract

Many industries keep log data on maintenance and support. The log data contains information on the problems reported as well as the actions taken to fix the problems. The log data contains a wealth of information useful for future maintenance, as well as product design and inventory management. However, it is always hard to identify parts involved automatically when the number of part types is large and the data are sloppily authored. Boeing’s P-MATCH identifies part names from noisy non-professionally authored log data. In this chapter, the authors use P-MATCH to illustrate how to leverage the combined strength of natural language processing and text mining.