LLMs to Support a Domain Specific Knowledge Assistant

TL;DR

This paper develops a domain-specific knowledge assistant for IFRS sustainability reporting by creating a synthetic QA dataset using LLMs and designing two question-answering architectures, achieving high accuracy and domain relevance.

Contribution

It introduces a novel generation and evaluation pipeline for creating a high-quality IFRS QA dataset and develops two effective QA architectures tailored for sustainability reporting.

Findings

QA dataset averages 8.16/10 on quality metrics

RAG pipeline achieves 85.32% accuracy on single-industry questions

LLM-based pipeline achieves 93.45% accuracy, outperforming baselines

Abstract

This work presents a custom approach to developing a domain specific knowledge assistant for sustainability reporting using the International Financial Reporting Standards (IFRS). In this domain, there is no publicly available question-answer dataset, which has impeded the development of a high-quality chatbot to support companies with IFRS reporting. The two key contributions of this project therefore are: (1) A high-quality synthetic question-answer (QA) dataset based on IFRS sustainability standards, created using a novel generation and evaluation pipeline leveraging Large Language Models (LLMs). This comprises 1,063 diverse QA pairs that address a wide spectrum of potential user queries in sustainability reporting. Various LLM-based techniques are employed to create the dataset, including chain-of-thought reasoning and few-shot prompting. A custom evaluation framework is developed to assess question and answer quality across multiple dimensions, including faithfulness, relevance, and domain specificity. The dataset averages a score range of 8.16 out of 10 on these metrics. (2) Two architectures for question-answering in the sustainability reporting domain - a RAG pipeline and a fully LLM-based pipeline. The architectures are developed by experimenting, fine-tuning, and training on the QA dataset. The final pipelines feature an LLM fine-tuned on domain specific data and an industry classification component to improve the handling of complex queries. The RAG architecture achieves an accuracy of 85.32% on single-industry and 72.15% on cross-industry multiple-choice questions, outperforming the baseline approach by 4.67 and 19.21 percentage points, respectively. The LLM-based pipeline achieves an accuracy of 93.45% on single-industry and 80.30% on cross-industry multiple-choice questions, an improvement of 12.80 and 27.36 percentage points over the baseline, respectively.