SRAuditor: An Automated Assessment Tool for Statement of Advice Documents

Abstract

Financial advice is given by a registered financial adviser (RFA) in the form of a statement of advice (SoA) document. To limit liability, financial advisor groups periodically assess SoA documents for compliance with legal regulations, internal policies, and best practices. However, this is a manual process that is often subjective, time-consuming and tedious. In this paper, we propose, implement and evaluate SoA Risk Auditor (SRAuditor), a natural language processing (NLP) framework to automatically assess and audit SoA documents. SRAuditor consists of two major components. The first one is a SoA transformer (SoA-T), a tool that automatically transforms and maps SoA document (generally a PDF). The other one is a question-answering engine (QA-R) that recommends legally compliant answers based on rule-based approaches for given SoA audit questions to assess and audit SoA documents. We validate the accuracy of SRAuditor's ability by evaluating it against assessments conducted by domain experts (i.e., financial advisors, lawyers). Experimental results using real-world SoA documents provided by our industry partner, Fourth Line Pty Limited indicate that SRAuditor has a high potential to be used for automatically assessing and auditing SoA documents.