Do LLMs Encode Frame Semantics? Evidence from Frame Identification

2025/09/23 by Chundru, Jayanth Krishna, Poddar, Rudrashis, Cao, Jie +1 · 2 citations
#Computation and Language (cs.CL) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.2509.19540

Abstract

We investigate whether large language models encode latent knowledge of frame semantics, focusing on frame identification, a core challenge in frame semantic parsing that involves selecting the appropriate semantic frame for a target word in context. Using the FrameNet lexical resource, we evaluate models under prompt-based inference and observe that they can perform frame identification effectively even without explicit supervision. To assess the impact of task-specific training, we fine-tune the model on FrameNet data, which substantially improves in-domain accuracy while generalizing well to out-of-domain benchmarks. Further analysis shows that the models can generate semantically coherent frame definitions, highlighting the model's internalized understanding of frame semantics.

Citations

• Potential and Limitations of LLMs in Capturing Structured Semantics: A Case Study on SRL
• Coarse-to-Fine Dual Encoders are Better Frame Identification Learners
• Few-Shot Semantic Parsing with Language Models Trained On Code
• LoRA: Low-Rank Adaptation of Large Language Models
• Learning Joint Semantic Parsers from Disjoint Data
• Frame-Semantic Parsing with Softmax-Margin Segmental RNNs and a Syntactic Scaffold

Cited by

• How Prompts Move Language Model Behavior: Frames, Salience, and Construal as Semantic Control
• Evaluating the Impact of LLM-Assisted Annotation in a Perspectivized Setting: the Case of FrameNet Annotation

Related

• Can LLMs Extract Frame-Semantic Arguments?
• Debugging Frame Semantic Role Labeling
• Exploring In-Context Learning for Frame-Semantic Parsing
• Knowledge-augmented Frame Semantic Parsing with Hybrid Prompt-tuning
• FrameEOL: Semantic Frame Induction using Causal Language Models