Synthetic Test Collections for Retrieval Evaluation

TL;DR

This paper explores the use of Large Language Models to generate fully synthetic test collections for information retrieval evaluation, including queries and relevance judgments, aiming to reduce costs and address data scarcity.

Contribution

It demonstrates that LLMs can reliably create synthetic test collections for IR evaluation, a novel approach that extends previous work on synthetic data generation.

Findings

Synthetic test collections can effectively evaluate IR systems.

LLMs can generate reliable synthetic relevance judgments.

Potential biases in LLM-generated collections need further investigation.

Abstract

Test collections play a vital role in evaluation of information retrieval (IR) systems. Obtaining a diverse set of user queries for test collection construction can be challenging, and acquiring relevance judgments, which indicate the appropriateness of retrieved documents to a query, is often costly and resource-intensive. Generating synthetic datasets using Large Language Models (LLMs) has recently gained significant attention in various applications. In IR, while previous work exploited the capabilities of LLMs to generate synthetic queries or documents to augment training data and improve the performance of ranking models, using LLMs for constructing synthetic test collections is relatively unexplored. Previous studies demonstrate that LLMs have the potential to generate synthetic relevance judgments for use in the evaluation of IR systems. In this paper, we comprehensively investigate whether it is possible to use LLMs to construct fully synthetic test collections by generating not only synthetic judgments but also synthetic queries. In particular, we analyse whether it is possible to construct reliable synthetic test collections and the potential risks of bias such test collections may exhibit towards LLM-based models. Our experiments indicate that using LLMs it is possible to construct synthetic test collections that can reliably be used for retrieval evaluation.