Deep parsing for the aviation industry: adjusting ARTEMIS for parsing simple clauses in ASD STE100

Abstract

The linguistically motivated parser ARTEMIS (Automatically Representing Text Meaning via an Interlingua-based System) has been designed [Periñán-Pascual and Arcas Túnez 2014] with the aim of providing formal representations of natural language fragments enriched with syntactic, semantic and pragmatic information. To test the validity of a parsing system before it is applied to natural language input it is a standard practice to apply it to a controlled language. In this paper we will address the treatment of simple clauses in ASD-STE 100, the controlled natural language employed by the European aviation industry for technical documentation. We will create or revise lexical and syntactic rules within ARTEMIS in order to make them sensitive to the formal impositions and the restricted communicative functions allowed in this technical language. These rules -which are consistent with the tenets of two linguistic models which substantiate ARTEMIS -namely, Role and Reference Grammar and the Lexical Constructional Model- will account for the first time for each of the different nodes that describe simple sentences in ARTEMIS, in attempt to make it suitable for the parsing of ASD-STE 100. We will conclude by indicating the areas where further research is needed for the full implementation of the parser.