GraphEval36K: Benchmarking Coding and Reasoning Capabilities of Large Language Models on Graph Datasets

TL;DR

This paper introduces GraphEval36K, a comprehensive benchmark dataset for evaluating large language models' capabilities in solving graph-related problems, and proposes Structured Symbolic Decomposition to enhance their performance.

Contribution

It provides the first extensive graph dataset for LLM evaluation and introduces SSD, a novel instruction-based method to improve LLM performance on graph tasks.

Findings

Private models outperform open-source models, but the gap is narrowing.

SSD improves LLM passing rates significantly across multiple models.

Performance varies across graph types and concepts, highlighting specific strengths and weaknesses.

Abstract

Large language models (LLMs) have achieved remarkable success in natural language processing (NLP), demonstrating significant capabilities in processing and understanding text data. However, recent studies have identified limitations in LLMs' ability to manipulate, program, and reason about structured data, especially graphs. We introduce GraphEval36K, the first comprehensive graph dataset, comprising 40 graph coding problems and 36,900 test cases to evaluate the ability of LLMs on graph problem-solving. Our dataset is categorized into eight primary and four sub-categories to ensure a thorough evaluation across different types of graphs. We benchmark ten LLMs, finding that private models outperform open-source ones, though the gap is narrowing. We also analyze the performance of LLMs across directed vs undirected graphs, different kinds of graph concepts, and network models. Furthermore, to improve the usability of our evaluation framework, we propose Structured Symbolic Decomposition (SSD), an instruction-based method designed to enhance LLM performance on complex graph tasks. Results show that SSD improves the average passing rate of GPT-4, GPT-4o, Gemini-Pro and Claude-3-Sonnet by 8.38%, 6.78%, 29.28% and 25.28%, respectively.