Each year, the air transportation industry faces more than 100,000 maintenance cases, including flight reports, maintenance orders for machine and equipment failures, and daily maintenance records. Such large amounts of structured and unstructured data are stored in the ERP system of the enterprise, which can be used as a reference for future case diagnosis and problem solving. However, different types of reports are written by different technicians and crew members, and may face spelling errors and word differences; and when the problem occurs, the technician must first understand what data to look for, where to gather the correct and useful information, and how to effectively utilize the collected data. It often takes a long time to find the right answer and may collect inaccurate and incomplete information.
In this case, in order to unlock the vast amount of information in maintenance records, through natural language processing (NLP) and advanced content analysis technology, the potential relevance information in the text can be excavated, so that technicians can quickly understand the causes of the problem and find appropriate solutions at the first time after the problem occurs, to ensure the safety of aircraft flight and maintain the smooth operation of the flight.