Explaining novel senses using definition generation with open language models

2025-09-30

AI Summary

This study addresses the problem of interpretable neologism definition generation in semantic change modeling. We propose a definition generation method leveraging open-weight large language models (LLMs), which takes contextual usage of a target word as input and outputs a natural-language definition. To our knowledge, this is the first work to apply open-weight LLMs to multilingual lexical semantic interpretation—specifically for Finnish, Russian, and German—within the AXOLOTL’24 shared task framework. We fine-tune both encoder-decoder and decoder-only architectures, achieving state-of-the-art performance over the best proprietary-model baseline. Empirical results demonstrate comparable effectiveness between the two architectures, confirming the viability and competitiveness of open-weight LLMs for this task. We publicly release a high-performance multilingual model, establishing a new paradigm and practical resource for interpretable semantic evolution research.

Technology Category

• Natural Language Processing: Lexical Semantics and Morphology
• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

We apply definition generators based on open-weights large language models to the task of creating explanations of novel senses, taking target word usages as an input. To this end, we employ the datasets from the AXOLOTL'24 shared task on explainable semantic change modeling, which features Finnish, Russian and German languages. We fine-tune and provide publicly the open-source models performing higher than the best submissions of the aforementioned shared task, which employed closed proprietary LLMs. In addition, we find that encoder-decoder definition generators perform on par with their decoder-only counterparts.

Problem

• Generating definitions for novel word senses
• Explaining semantic changes in multiple languages
• Comparing open-source and closed LLM performance

Innovation

• Using open-weights LLMs for definition generation
• Fine-tuning models outperforming proprietary LLM submissions
• Encoder-decoder generators matching decoder-only model performance