Leveraging Lora Fine-Tuning and Knowledge Bases for Construction Identification

TL;DR

This paper presents a method combining LoRA fine-tuning and knowledge bases to improve the automatic identification of the English ditransitive construction, demonstrating significant performance gains over baseline models.

Contribution

It introduces a novel approach integrating LoRA fine-tuning with RAG for construction identification, showing improved semantic understanding over pattern matching.

Findings

LoRA fine-tuning outperforms baseline models

Semantic understanding improves with fine-tuning

Model achieves higher accuracy on annotated corpus

Abstract

This study investigates the automatic identification of the English ditransitive construction by integrating LoRA-based fine-tuning of a large language model with a Retrieval-Augmented Generation (RAG) framework.A binary classification task was conducted on annotated data from the British National Corpus. Results demonstrate that a LoRA-fine-tuned Qwen3-8B model significantly outperformed both a native Qwen3-MAX model and a theory-only RAG system. Detailed error analysis reveals that fine-tuning shifts the model's judgment from a surface-form pattern matching towards a more semantically grounded understanding based.