Knowledge base question answering for space debris queries

Darm, Paul and Miceli-Barone, Antonio Valerio and Cohen, Shay B. and Riccardi, Annalisa; (2023) Knowledge base question answering for space debris queries. In: Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics. Association for Computational Linguistics, Toronto, pp. 487-499. ISBN 9781959429685 (https://aclanthology.org/2023.acl-industry.0)

[thumbnail of Darm-etal-ACL2023-Knowledge-base-question-answering-for-space-debris-queries]
Preview
 Text. Filename: Darm_etal_ACL2023_Knowledge_base_question_answering_for_space_debris_queries.pdf
Final Published Version
License: Creative Commons Attribution 4.0 logo
Download (319kB)| Preview

Abstract

Space agencies execute complex satellite operations that need to be supported by the technical knowledge contained in their extensive information systems. Knowledge bases (KB) are an effective way of storing and accessing such information at scale. In this work we present a system, developed for the Euro-pean Space Agency (ESA), that can answer complex natural language queries, to support engineers in accessing the information contained in a KB that models the orbital space debris environment. Our system is based on a pipeline which first generates a sequence of basic database operations, called a sketch, from a natural language question, then specializes the sketch into a concrete query program with mentions of entities, attributes and relations, and finally executes the program against the database. This pipeline decomposition approach enables us to train the system by lever-aging out-of-domain data and semi-synthetic data generated by GPT-3, thus reducing over-fitting and shortcut learning even with limited amount of in-domain training data. Our code can be found at https://github.com/PaulDrm/DISCOSQA.

CORE (COnnecting REpositories)