An Integrated Retrieval-Augmented Generation Approach for Tax Law Assistance and Decision Support

DOI

10.2991/978-94-6463-948-3_3How to use a DOI?

Keywords

Retrieval-Augmented Generation (RAG); Large Language Models (LLMs); FAISS Vector Indexing; T5 Transformer; Multilingual NLP; TF-IDF Vectorization; LLAMA3; Decision Support Systems; Document Summarization; Tax Law Assistance

Abstract

Tax laws in India are complex for salaried employees and small businesses, frequently altered by disparate information sources, and require familiarity with specialist knowledge areas to understand legal text. In this paper, we suggest an AI-augmented Tax Law Assis- tant with Retrieval-Augmented Generation, multilingual understanding, and personalized recommendation modules to ease tax compliance and financial planning. Additionally, the system can analyze wide-ranging le- gal guidelines and condense them into concise, context-rich bullet points, as well as create custom tax-efficient investment portfolios based on user profiles, employment data, objectives, and risk tolerance. The platform runs an LLAMA3 model on Ollama to generate context, FAISS to se- mantically search the Income Tax Act, and a fine-tuned T5 transformer to accurately translate the response to English-Hindi. The tax law query answer’s evaluation showed that the system achieved 54.9% fully accurate response, 21.5% partially accurate response, and an overall accuracy of 76.4% for tax law queries. The translation module, with a BLEU score of 76, is said to have good bilingual reliability. Latencies of different LLMs show that LLAMA3 which responds in 12.47s on an average, provides the best quality response, but Phi3 has the least latency at an aver- age of 7.93s. When compared with software such as PolicyBazaar and ClearTax, the proposed open-source tool provides full personalization and multilingualism and documents processing.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Sonal Gholap
AU  - Omkar Bhoir
AU  - Pradnyesh Jain
AU  - Sunil Ghane
PY  - 2026
DA  - 2026/01/06
TI  - An Integrated Retrieval-Augmented Generation Approach for Tax Law Assistance and Decision Support
BT  - Proceedings of the International Conference on Sustainable Innovation with Artificial Intelligence and Machine Learning 2025 (ICSIAIML 2025)
PB  - Atlantis Press
SP  - 16
EP  - 29
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6463-948-3_3
DO  - 10.2991/978-94-6463-948-3_3
ID  - Gholap2026
ER  -