Multilingual Extraction and Categorization of Lexical Collocations with Graph-aware Transformers
This work addresses a challenging task for language learning, dictionary compilation, and NLP applications, but it is incremental as it builds on existing BERT models with architectural enhancements.
The paper tackled the problem of recognizing and categorizing lexical collocations in context across multiple languages by proposing a BERT-based model enhanced with a graph-aware transformer architecture, achieving results that show explicit encoding of syntactic dependencies improves performance and provides insights into collocation typification in English, Spanish, and French.
Recognizing and categorizing lexical collocations in context is useful for language learning, dictionary compilation and downstream NLP. However, it is a challenging task due to the varying degrees of frozenness lexical collocations exhibit. In this paper, we put forward a sequence tagging BERT-based model enhanced with a graph-aware transformer architecture, which we evaluate on the task of collocation recognition in context. Our results suggest that explicitly encoding syntactic dependencies in the model architecture is helpful, and provide insights on differences in collocation typification in English, Spanish and French.