ANALOGICAL -- A Novel Benchmark for Long Text Analogy Evaluation in Large Language Models

TL;DR

This paper introduces ANALOGICAL, a benchmark for evaluating large language models' ability to understand and draw analogies across long texts with increasing complexity, revealing current limitations.

Contribution

It presents a new benchmark for intrinsic evaluation of LLMs on long text analogies across six complexity levels, filling a gap in existing evaluation methods.

Findings

LLMs struggle more with higher levels of analogy complexity

The benchmark uses 13 datasets and 3 distance measures

Evaluation shows decreasing performance as analogy complexity increases

Abstract

Over the past decade, analogies, in the form of word-level analogies, have played a significant role as an intrinsic measure of evaluating the quality of word embedding methods such as word2vec. Modern large language models (LLMs), however, are primarily evaluated on extrinsic measures based on benchmarks such as GLUE and SuperGLUE, and there are only a few investigations on whether LLMs can draw analogies between long texts. In this paper, we present ANALOGICAL, a new benchmark to intrinsically evaluate LLMs across a taxonomy of analogies of long text with six levels of complexity -- (i) word, (ii) word vs. sentence, (iii) syntactic, (iv) negation, (v) entailment, and (vi) metaphor. Using thirteen datasets and three different distance measures, we evaluate the abilities of eight LLMs in identifying analogical pairs in the semantic vector space. Our evaluation finds that it is increasingly challenging for LLMs to identify analogies when going up the analogy taxonomy.