UniSumEval: Towards Unified, Fine-Grained, Multi-Dimensional Summarization Evaluation for LLMs

TL;DR

UniSumEval introduces a comprehensive benchmark for summarization evaluation that covers diverse input scenarios and multiple evaluation dimensions, utilizing AI assistance for data creation and annotation.

Contribution

It presents a new benchmark with fine-grained, multi-dimensional annotations across varied input contexts, and benchmarks recent language models and evaluation methods.

Findings

Insights into model performance across different contexts and dimensions

Comparison of state-of-the-art automated evaluators

Enhanced annotation quality with AI assistance

Abstract

Existing benchmarks for summarization quality evaluation often lack diverse input scenarios, focus on narrowly defined dimensions (e.g., faithfulness), and struggle with subjective and coarse-grained annotation schemes. To address these shortcomings, we create UniSumEval benchmark, which extends the range of input context (e.g., domain, length) and provides fine-grained, multi-dimensional annotations. We use AI assistance in data creation, identifying potentially hallucinogenic input texts, and also helping human annotators reduce the difficulty of fine-grained annotation tasks. With UniSumEval, we benchmark nine latest language models as summarizers, offering insights into their performance across varying input contexts and evaluation dimensions. Furthermore, we conduct a thorough comparison of SOTA automated summary evaluators. Our benchmark data will be available at https://github.com/DISL-Lab/UniSumEval-v1.0.