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

2026-04-15 · Source: Computation and Language · Field: Technology & Digital — Artificial Intelligence & Machine Learning · Depth: Expert, quick

Summary

A new method for automatically expanding WordNet-style lexical resources to new languages has been developed, focusing on generating senses by associating target-language lemmas with existing lexical concepts through semantic projection. The approach utilizes a sense-tagged English corpus and its translation, projecting English synsets onto aligned target-language tokens and assigning corresponding lemmas to those synsets. To enhance alignment quality, a pre-trained base aligner is augmented with a bilingual dictionary, which also serves to filter out incorrect sense projections. This project-and-filter strategy has been evaluated across multiple languages, demonstrating improved precision compared to prior methods, dictionary-based, and large language model baselines, while maintaining interpretability and requiring minimal external resources. The code, documentation, and generated sense inventories are planned for public release.

Key takeaway

For research scientists working on multilingual natural language processing or lexical resource creation, this dictionary-based cross-lingual sense projection method offers a precise and interpretable way to expand WordNet-style resources. You should consider integrating this project-and-filter strategy to improve the quality of your generated sense inventories, especially given its low external resource requirements.

Key insights

A dictionary-augmented projection method improves cross-lingual WordNet expansion precision with minimal resources.

Principles

• Semantic projection links lemmas to synsets.
• Bilingual dictionaries enhance alignment quality.
• Filtering incorrect projections improves precision.

Method

The method projects English synsets onto aligned target-language tokens from a translated corpus, assigning lemmas. A bilingual dictionary augments alignment and filters projections to ensure quality.

In practice

• Expand WordNet-style resources.
• Generate cross-lingual sense inventories.

Topics

• WordNet
• Lexical Resources
• Sense Generation
• Cross-Lingual Projection
• Bilingual Dictionary

Best for: Research Scientist, AI Scientist, NLP Engineer

Related on AIssential

• Source-Grounded Semantic Reinforcement Learning for Low-Resource Target-Language Generation — SG-SRL uses source-language monolingual data with cross-lingual semantic RL to improve low-resource target-language generation, recovering fluency with a small parallel corpus.
• Towards Building a Computational Sense Inventory from the Monier-Williams Dictionary Using Clustering Techniques — Computational sense inventories for Sanskrit can be built from the Monier-Williams Dictionary using clustering techniques.
• Just Use XML: Revisiting Joint Translation and Label Projection — Joint translation and label projection with XML tags can improve both translation quality and label transfer.
• Giving Voice to the Constitution: Low-Resource Text-to-Speech for Quechua and Spanish Using a Bilingual Legal Corpus — Cross-lingual transfer in TTS effectively mitigates data scarcity for low-resource languages like Quechua.
• Scene Abstraction for Lexical Semantics: Structured Representations of Situated Meaning — "Scene Abstraction" provides a structured framework to represent the situated, interpretive dimensions of lexical meaning using LLMs.
• CLP-Transfer: Cross-Lingual and Progressive Transfer Learning — CLP-Transfer simplifies cross-lingual model transfer using token overlap and a small helper model, but shows limited downstream task performance.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computation and Language.