An Investigation of LLMs' Inefficacy in Understanding Converse Relations

TL;DR

This paper evaluates whether large language models truly understand structured semantics by testing them on a new benchmark for converse relations, revealing limitations and tendencies for shortcut learning.

Contribution

Introduces ConvRe, a novel benchmark for assessing LLMs' understanding of converse relations in formal language, with comprehensive evaluation protocols and analysis.

Findings

LLMs show limited understanding of converse relations.

Models tend to rely on shortcut learning rather than genuine comprehension.

Scaling improves performance but does not fully solve the understanding challenge.

Abstract

Large Language Models (LLMs) have achieved remarkable success in many formal language oriented tasks, such as structural data-to-text and semantic parsing. However current benchmarks mostly follow the data distribution of the pre-training data of LLMs. Therefore, a natural question rises that do LLMs really understand the structured semantics of formal languages. In this paper, we investigate this problem on a special case, converse binary relation. We introduce a new benchmark ConvRe focusing on converse relations, which contains 17 relations and 1240 triples extracted from popular knowledge graph completion datasets. Our ConvRE features two tasks, Re2Text and Text2Re, which are formulated as multi-choice question answering to evaluate LLMs' ability to determine the matching between relations and associated text. For the evaluation protocol, apart from different prompting methods, we further introduce variants to the test text and few-shot example text. We conduct experiments on three popular LLM families and have observed various scaling trends. The results suggest that LLMs often resort to shortcut learning and still face challenges on our proposed benchmark.