Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Citation-Enforced RAG for Fiscal Document Intelligence: Cited, Explainable Knowledge Retrieval in Tax Compliance

Akhil Chandra Shanivendra

Abstract

Tax authorities and public-sector financial agencies rely on large volumes of unstructured and semi-structured fiscal documents - including tax forms, instructions, publications, and jurisdiction-specific guidance - to support compliance analysis and audit workflows. While recent advances in generative AI and retrieval-augmented generation (RAG) have shown promise for document-centric question answering, existing approaches often lack the transparency, citation fidelity, and conservative behaviour required in high-stakes regulatory domains. This paper presents a multimodal, citation-enforced RAG framework for fiscal document intelligence that prioritises explainability and auditability. The framework adopts a source-first ingestion strategy, preserves page-level provenance, enforces citations during generation, and supports abstention when evidence is insufficient. Evaluation on real IRS and state tax documents demonstrates improved citation fidelity, reduced hallucination, and analyst-usable explanations, illustrating a pathway toward trustworthy AI for tax compliance.

Citation-Enforced RAG for Fiscal Document Intelligence: Cited, Explainable Knowledge Retrieval in Tax Compliance

Abstract

Tax authorities and public-sector financial agencies rely on large volumes of unstructured and semi-structured fiscal documents - including tax forms, instructions, publications, and jurisdiction-specific guidance - to support compliance analysis and audit workflows. While recent advances in generative AI and retrieval-augmented generation (RAG) have shown promise for document-centric question answering, existing approaches often lack the transparency, citation fidelity, and conservative behaviour required in high-stakes regulatory domains. This paper presents a multimodal, citation-enforced RAG framework for fiscal document intelligence that prioritises explainability and auditability. The framework adopts a source-first ingestion strategy, preserves page-level provenance, enforces citations during generation, and supports abstention when evidence is insufficient. Evaluation on real IRS and state tax documents demonstrates improved citation fidelity, reduced hallucination, and analyst-usable explanations, illustrating a pathway toward trustworthy AI for tax compliance.
Paper Structure (57 sections, 3 figures, 2 tables)

This paper contains 57 sections, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Background and Motivation
  3. Problem Statement
  4. Contributions
  5. Related Work
  6. Document Intelligence in Public-Sector and Financial Domains
  7. Retrieval-Augmented Generation
  8. Multimodal RAG and Document-Centric Systems
  9. Explainable and Trustworthy AI
  10. Positioning of This Work
  11. Dataset Collection and Corpus Characteristics
  12. Data Sources
  13. Corpus Composition and Statistics
  14. Multimodal Document Characteristics
  15. Ethical and Legal Considerations
  16. ...and 42 more sections

Figures (3)

  • Figure 1: High-level architecture of the Source-First Multimodal RAG framework. Unlike standard approaches that summarise content before embedding, this system embeds raw text segments linked to specific document identifiers and page ranges to enable downstream citation enforcement.
  • Figure 2: Example of citation-enforced answer generation. The system retrieves relevant chunks from California state guidance and synthesises an answer where every claim is explicitly linked to a source document and page number.
  • Figure 3: Flowchart showing the decision logic for Similarity Checks and Citation Validation.