ATC-ANNO: Semantic Annotation for Air Traffic Control with Assistive Auto-Annotation

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 6375–6380 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 6375 ATC-ANNO: Semantic Annotation for Air Traffic Control with Assistive Auto-Annotation Marc Schulder*†, Johannah O’Mahony*, Yury Bakanouski*, Dietrich Klakow* * Spoken Language Systems, Saarland University, Germany † Institute for German Sign Language, University of Hamburg, Germany marc.schulder@uni-hamburg.de, omahony@coli.uni-saarland.de, yury.bakanouski@gmail.com, dietrich.klakow@lsv.uni-saarland.de Abstract In air traffic control, assistant systems support air traffic controllers in their work. To improve the reactivity and accuracy of the assistant, automatic speech recognition can monitor the commands uttered by the controller. However, to provide sufficient training data for the speech recognition system, many hours of air traffic communications have to be transcribed and semantically annotated. For this purpose we develop the annotation tool ATC-ANNO. It provides a number of features to support the annotator in their task, such as auto-complete suggestions for semantic tags, access to preliminary speech recognition predictions, syntax highlighting and consistency indicators. Its core assistive feature, however, is its ability to automatically generate semantic annotations. Although it is based on a simple hand-written finite state grammar, it is also able to annotate sentences that deviate from this grammar. We evaluate the impact of different fea- tures on annotator efficiency and find that automatic annotation allows annotators to cover four times as many utterances in the same time. Keywords: Annotation Tool; Controlled Languages; Air Traffic Control; Assisted Annotation 1. Introduction Air traffic control (ATC) communications is a challenging domain for automatic speech recognition (ASR) (Hamel et al., 1989; Cordero et al., 2012; Shore et al., 2012). Very little training data for the ATC domain is publicly available, necessitating the recording, transcription and annotation of more data to allow ASR to be successfully integrated into ATC systems (Oualil et al., 2015). We present ATC-ANNO, a tool for the transcription and se- mantic annotation of recordings of air traffic control com- munications. Transcribing ATC communications is chal- lenging, as air traffic controllers speak extremely fast, due to the time-sensitive nature of their task. In addition they are usually non-native speakers and their utterances follow a specific information-dense phraseology. Due to these challenges, annotators must have prior experi- ence with ATC communications. Unfortunately, this means that appropriate candidates usually have no experience in transcription or annotation of natural language data. They are domain experts, but novice annotators. We therefore try to support them particularly in more technical tasks, like writing XML tags for the semantic layer of the annotation, while keeping the interface as focussed as possible. ATC-ANNO provides a number of assistive features. These range from basic features, such as syntax highlighting and consistency checks between different annotation layers, to more immediate support, like XML tag auto-completion suggestions or access to preliminary speech recognition predictions. Our most advanced assistive feature is the automatic generation of semantic annotations, based on a grammar of common phrases encountered in air traffic con- trol. While the grammar is written as a finite state automa- …

Content truncated to 500 words