Universal Information Extraction (UIE) for Financial Documents Analysis and ESG Intelligence (ART/364CP)

Universal Information Extraction (UIE) for Financial Documents Analysis and ESG Intelligence (ART/364CP)

Universal Information Extraction (UIE) for Financial Documents Analysis and ESG Intelligence (ART/364CP)
ART/364CP
Platform
23 / 05 / 2023 - 31 / 08 / 2024
6,969.000

Dr Tao YU

Bank of China (Hong Kong) Limited
Deloitte Advisory (Hong Kong) Limited


This project aims to build an AI analytics engine for diversified unstructured data processing, under small labeling data sets and different information extraction targets. The engine adopts Universal Information Extraction (UIE) and Few-shot learning technologies to achieve high efficiency, scalability, and generalizability. The project will develop few-shot learning models for document matching to support banking loan document review, as well as UIE-based information extraction models to get ESG insights from ESG reports and support applications of • Document summary and compliance checking • Company ESG profiling and improvement suggestions • Industry ESG benchmark and peer comparison The developed solution can serve financial institutes, investors, regulators, and other parties. It can also be easily extended to handle other types of unstructured data analysis tasks.