Explainable Information Retrieval in the Audit Domain

2025-07-04

AI Summary

Audit is a high-risk, highly specialized domain where explainable information retrieval (XIR) remains unexplored despite critical needs—such as mitigating hallucinated citations in conversational systems and ensuring audit decisions are traceable, evidence-based, and accountable. Method: This work introduces the first XIR framework tailored for auditing tasks, grounded in core design principles of result traceability, evidence verifiability, and reasoning interpretability, and explicitly integrated with domain-specific auditing knowledge. It systematically analyzes unique XIR requirements and challenges in auditing—including regulatory compliance, heterogeneous evidence integration, and unambiguous responsibility attribution—and proposes domain-adapted evaluation dimensions and a technical roadmap for XIR advancement. Contribution/Results: The framework establishes a theoretical foundation and practical paradigm for developing trustworthy, robust intelligent auditing tools, bridging a critical gap between XIR research and real-world audit practice.

Technology Category

• Humans and AI: Explainable AI (XAI) for Human Understanding
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Philosophy and Ethics of AI: Accountability, Interpretability & Explainability

Application Category

• Responsible Web: Audits of web technologies, standards, platforms, and applications
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity

Abstract

Conversational agents such as Microsoft Copilot and Google Gemini assist users with complex search tasks but often generate misleading or fabricated references. This undermines trust, particularly in high-stakes domains such as medicine and finance. Explainable information retrieval (XIR) aims to address this by making search results more transparent and interpretable. While most XIR research is domain-agnostic, this paper focuses on auditing -- a critical yet underexplored area. We argue that XIR systems can support auditors in completing their complex task. We outline key challenges and future research directions to advance XIR in this domain.

Problem

• Address misleading references in conversational search agents
• Enhance transparency in audit-domain information retrieval
• Advance explainable retrieval for high-stakes auditing tasks

Innovation

• Focuses on Explainable Information Retrieval in auditing
• Makes search results transparent and interpretable
• Addresses challenges in high-stakes audit domain