Content Generation for Information Retrieval Systems: A Survey from a Supply-Side Optimization Perspective

Abstract

With the rapid growth of commercial content platforms (e.g., TikTok, YouTube), modern information retrieval (IR) systems have evolved into two-sided ecosystems with both supply sides (e.g., content creators) and demand sides (e.g., users). From an economic perspective, traditional IR research primarily focuses on optimizing supply-demand matching through IR models while treating content supply as exogenous and fixed. However, such a design may pose risks of market failure in two-sided platforms, leading to degraded user experience and severe Matthew effects. Recently, content generation techniques powered by generative models have emerged as a potential solution to such risks by enabling high-quality and scalable supply-side optimization. However, this development also introduces new challenges to IR systems, including difficulties in evaluation and potential risks to the content ecosystem. In this paper, we provide a comprehensive survey of recent advances in content generation for IR systems from the perspective of economic supply-side optimization. Our review is structured along three key dimensions: methods, evaluation, and challenges. Specifically, grounded in a two-sided market framework, we categorize existing content generation approaches into two broad classes according to their demand-side optimization objectives: generic content generation and personalized content generation. Next, we systematically review evaluation strategies for content generation in IR systems, progressing from high-level objectives and methodological paradigms to fine-grained dimensions and associated metrics. Finally, we identify key challenges and highlight promising future directions, with the goal of advancing the understanding and development of content generation for supply-side optimization in IR systems.

Keywords

ontent generation / information retrieval / generative model

RIGHTS & PERMISSIONS

Higher Education Press 2026

Just Accepted

This article has successfully passed peer review and final editorial review, and will soon enter typesetting, proofreading and other publishing processes. The currently displayed version is the accepted final manuscript. The officially published version will be updated with format, DOI and citation information upon launch. We recommend that you pay attention to subsequent journal notifications and preferentially cite the officially published version. Thank you for your support and cooperation.