Improving Zero-Shot Text Matching for Financial Auditing with Large Language Models
Authors:
- Lars HillebrandLHLars Hillebrand
- Armin BergerABArmin Berger
- Tobias DeußerTDTobias Deußer
- Tim DilmaghaniTDTim Dilmaghani
- Mohamed KhaledMKMohamed Khaled
- Bernd KliemBKBernd Kliem
- Rüdiger LoitzRLRüdiger Loitz
- Maren PielkaMPMaren Pielka
- David LeonhardDLDavid Leonhard
- Christian BauckhageCBChristian Bauckhage
- Rafet SifaRSRafet Sifa
Open Access
- 22 August 2023
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Abstract
Auditing financial documents is a very tedious and time-consuming process. As of today, it can already be simplified by employing AI-based solutions to recommend relevant text passages from a report for each legal requirement of rigorous accounting standards. However, these methods need to be fine-tuned regularly, and they require abundant annotated data, which is often lacking in industrial environments. Hence, we present ZeroShotALI, a novel recommender system that leverages a state-of-the-art large language model (LLM) in conjunction with a domain-specifically optimized transformer-based text-matching solution. We find that a two-step approach of first retrieving a number of best matching document sections per legal requirement with a custom BERT-based model and second filtering these selections using an LLM yields significant performance improvements over existing approaches.Keywords
All Related Versions
This publication has 7 references indexed in Scilit:
- Automatic Consistency Checking of Table and Text in Financial DocumentsProceedings of the Northern Lights Deep Learning Workshop, 2023
- Towards automating Numerical Consistency Checks in Financial Reports2022
- Zero-Shot Text Matching for Automated Auditing using Sentence Transformers2022
- KPI-BERT: A Joint Named Entity Recognition and Relation Extraction Model for Financial Reports2022
- ALiBERT2021
- Towards Automated Auditing with Machine Learning2019
- Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks2019