Airbus AI researchers have developed a system that uses natural language understanding to improve question answering (QA) performance when flight crews search for aircraft operating information.
The aerospace industry relies on technical documents such as Aircraft Operating Manuals (AOM), Aircraft Operating Instructions and particularly Flight Crew Operating Manuals (FCOM) to guide flight crews on aircraft operations under normal, abnormal, and emergency conditions. FCOMs are issued by aircraft manufacturers and cover system descriptions, procedures, techniques, and performance data. They are the references used to develop standard operating procedures to improve safety and efficiency.
Most government aviation administrations have authorized the use of tablet computers by commercial carrier pilots and flight crews to access FCOM information. The Airbus AI researchers note however that existing electronic flight bag (EFB) systems used for this purpose are in practice little more than pdf viewers with keyword search functionality. The new Smart Librarian (SL) system instead uses natural language understanding and interactive search to boost QA performance on FCOMs.
Airbus AI’s SL FCOM QA system has three main components: a dialogue engine, a retriever (search engine), and a QA module.
The system uses the natural language-based, open-source Rasa framework as its dialogue engine, enabling it to recognize high-level user utterances, predict the next best action, and identify users’ questions based on dialogue state. The retriever system was built under a classic QA system architecture, with an overall documents filter function designed to source answers within a limited timeframe. The retriever also leverages the probabilistic relevance framework BM25 as an indexing scheme to enhance performance.
With the dialogue engine and retriever filter search system in place, the researchers’ next challenge was triggering the QA modules to give a correct and final answer in natural language. To do this, they leveraged large Google BERT (Bidirectional Encoder Representations from Transformers) models, a DrQA model, a multi-task Learning approach and an XGBoost.
In interactive user experiments, the SL system outperformed the EFB’s FCOM keyword search functionality, finding answers with higher accuracy and in less time either with or without the inclusion of keywords in the queries.
The Airbus AI team identifies future areas for improvement to the SL system as increasing the amount of in-domain data to enable further fine-tuning of the QA engine, as well as improving the system’s ability to extract answers from FCOM tables.
The paper A Question-Answering System for Aircraft Pilots’ Documentation is on arXiv.
Analyst: Robert Tian | Editor: Michael Sarazen & Fangyu Cai
This report offers a look at how China has leveraged artificial intelligence technologies in the battle against COVID-19. It is also available on Amazon Kindle. Along with this report, we also introduced a database covering additional 1428 artificial intelligence solutions from 12 pandemic scenarios.
Click here to find more reports from us.
We know you don’t want to miss any news or research breakthroughs. Subscribe to our popular newsletter Synced Global AI Weekly to get weekly AI updates.