Multi-layer semantic interpretation of Aeronautical information

Gérard M Dupont

Abstract: We present a hybrid architecture for knowledge extraction, in our future digital assistant for pilots in commercial aircraft. We need to automatically decode aeronautical messages known as NOTAM (Notice to Airmen) and ATIS (Automatic terminal area information service). Those messages include natural language sentences for operationally relevant information, particularly about the availability of runways or taxiways, related limitations or procedures. For example, ILS APCHS IN USE RWY 36R, RNAV RWY 36L APCHS IN USE (“APCHS” stands for approaches, “RWY” stands for runway). Such messages need to be reliably interpreted into a logical form, in order to: push relevant notifications to the pilot, answer his questions, or monitor flight’s compliance with applicable constraints. Moreover, in case of pilot incapacitation during a single pilot flight, the aircraft should be able to autonomously divert the flight.