CypherBench

该数据集名为CypherBench，包含11个大规模、跨领域的属性图，拥有780万个实体和超过1万个问题，旨在评估在大型语言模型（LLMs）背景下图形检索的性能。此外，该数据集还支持评估各种最先进的LLMs在零样本文本到Cypher查询性能方面的表现。其规模之大，实体数量达到780万，任务重点在于图形检索和Cypher查询生成。

This dataset, CypherBench, contains 11 large-scale, cross-domain property graphs totaling 7.8 million entities and over 10,000 questions. It is intended to assess the performance of graph retrieval in the context of large language models (LLMs). Furthermore, this dataset enables evaluation of various state-of-the-art LLMs on their zero-shot text-to-Cypher query generation capabilities. Boasting a substantial scale, its core tasks focus on graph retrieval and Cypher query generation.

搜集汇总

背景与挑战

背景概述

CypherBench是一个大规模、跨领域的属性图数据集，包含11个图、780万个实体和超过1万个问题，主要用于评估大型语言模型在图形检索和零样本文本到Cypher查询生成方面的性能。该数据集以其庞大的实体数量和专注于图形检索任务的特点，为LLMs的性能测试提供了全面基准。