What question did this study set out to answer?

This research aims to evaluate classification strategies for detecting argument temporal references using smaller language models.

synapse

⌘+K

synapse

⌘+K

April 1, 2026Open Access

AIDAVANCE at the NTCIR-18 FinArg-2 Task: Making the Most of Small Language Models

Key Points

This research aims to evaluate classification strategies for detecting argument temporal references using smaller language models.
Utilized various classification strategies, including multi-class classification and a hierarchical cascade approach.
Employed an LLM-based argument rewriting method.
Fine-tuned the mDeBERTa model for performance assessment.
Achieved fourth place overall in the task rankings.
Reported a Micro-F1 score of 0.6905 and a Macro-F1 score of 0.6711.
Demonstrated that fine-tuning smaller models can outperform larger state-of-the-art LLMs.

Abstract

This paper presents AIDAVANCE's approach to Subtask 2 (Detection of Argument Temporal References) of the NTCIR-18 FinArg-2 Task. We explored different classification strategies, including direct multi-class classification, a hierarchical cascade approach that first identifies the presence of a temporal reference before further categorization, and an LLM-based argument rewriting method. Our best model, a fine-tuned mDeBERTa using the multi-class approach, ranked fourth overall, achieving a Micro-F1 score of 0.6905 and a Macro-F1 score of 0.6711. Our findings reinforce that fine-tuning smaller encoder models remains an effective strategy for specialized classification tasks, even outperforming state-of-the-art LLMs.

Read Full Paperexternally

Mark Helpful

Bookmark

Relay

View Full Paper