Generating Concept Lexicalizations via Dictionary-Based Cross-Lingual Sense Projection

The paper presents a method for expanding WordNet-style lexical resources to new languages through a process of sense generation via dictionary-based cross-lingual sense projection. The approach enhances a pretrained aligner with a bilingual dictionary to improve alignment quality and employs a project-and-filter strategy, demonstrating improved precision over existing methods. The code and generated sense inventories are publicly available, which is significant for practitioners aiming to enhance multilingual NLP applications with interpretable and resource-efficient techniques.