Extracting Space Situational Awareness Events from News Text

TL;DR

This paper introduces a neural extraction system for space situational awareness that uses news text to identify key events like launches and failures, achieving high accuracy in a low-resource domain.

Contribution

It presents a novel corpus of news articles and a dependency-rule-based extraction system for space events, advancing the use of textual data in space situational awareness.

Findings

Achieved an F1 score between 53 and 91 per slot for event extraction.

Constructed a large corpus of 48.5k news articles from 2009-2020.

Demonstrated the effectiveness of neural models in low-resource space event extraction.

Abstract

Space situational awareness typically makes use of physical measurements from radar, telescopes, and other assets to monitor satellites and other spacecraft for operational, navigational, and defense purposes. In this work we explore using textual input for the space situational awareness task. We construct a corpus of 48.5k news articles spanning all known active satellites between 2009 and 2020. Using a dependency-rule-based extraction system designed to target three high-impact events -- spacecraft launches, failures, and decommissionings, we identify 1,787 space-event sentences that are then annotated by humans with 15.9k labels for event slots. We empirically demonstrate a state-of-the-art neural extraction system achieves an overall F1 between 53 and 91 per slot for event extraction in this low-resource, high-impact domain.