Handling Non-compositionality in Multilingual CNLs

Abstract

Abstract. In this paper, we describe methods for handling multilingual noncompositional constructions in the framework of GF. We specifically look at methods to detect and extract non-compositional phrases from parallel texts and propose methods to handle such constructions in GF grammars. We expect that the methods to handle non-compositional constructions will enrich CNLs by providing more flexibility in the design of controlled languages. We look at two specific use cases of non-compositional constructions: a general-purpose method to detect and extract multilingual multiword expressions and a procedure to identify nominal compounds in German. We evaluate our procedure for multiword expressions by performing a qualitative analysis of the results. For the experiments on nominal compounds, we incorporate the detected compounds in a full SMT pipeline and evaluate the impact of our method in machine translation process.

Compositional matrix-space models of language: Definitions, properties, and learning methods

“…In the first part of the experiments, we investigate our learning method for CMSMs on the task of compositionality prediction of multi-word expressions (MWE). Compositionality prediction is important in downstream NLP tasks such as statistical machine translation (Enache, Listenmaa, and Kolachina 2014;Weller et al 2014), word-sense disambiguation (McCarthy, Keller, and Carroll 2003), and text summarization (ShafieiBavani et al 2018) where a method is required to detect whether the words in a phrase are used in a compositional meaning. Therefore, we choose to evaluate the proposed method for CMSMs on the ability to detect the compositionality of phrases.…”