Document Classification Modeling for Fuzzy Audit Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional business process audits rely heavily on the general ledger as the primary source of truth, which leaves 80% of the data untapped and prone to errors, leading to incomplete audits and inefficient data collection processes.
Innovation Solution
A data-driven process model is used to align and classify data from various sources, determining the likelihood of data alignment with the ground truth and flagging potential sources of information for verification or disambiguation, employing fuzzy logic to handle imperfect data and automate the audit process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the general ledger is used as the primary source of truth, then the audit process is simplified and straightforward, but data accuracy and completeness deteriorate due to errors, corruption, and missing entries
Solution Approach 1:
The patent introduces an intermediary system that sits between the general ledger and the audit process. This system retrieves original source documents (invoices, contracts, emails) and uses AI/ML models to extract and verify data, acting as a mediator that preserves ledger simplicity while ensuring data accuracy through independent verification
Solution Approach 2:
The system performs preliminary actions by proactively retrieving and validating source documents before the audit process begins. It continuously monitors and verifies data integrity by comparing ledger entries against original sources, preventing errors from propagating through the audit process
2Productivity
If only the general ledger is audited, then the audit scope is limited and focused, but 80% of available data remains untapped and unverified
Solution Approach 1:
The patent creates a universal audit system that can handle multiple data types and sources simultaneously. The AI/ML models are designed to process structured ledger data, unstructured source documents, emails, contracts, and invoices within a single framework, enabling comprehensive verification across all 80% previously untapped data
Solution Approach 2:
The system adds another dimension to the audit process by moving from a single-layer ledger audit to a multi-layer verification approach. It creates a new dimension of data retrieval and validation by accessing original source documents and supporting evidence that exist outside the traditional ledger structure
3Reliability
If manual data collection and verification is performed, then data accuracy can be ensured, but time consumption and costs increase significantly
Solution Approach 1:
The patent replaces manual mechanical verification processes with automated AI/ML systems. Machine learning models automatically extract data from source documents, verify consistency with ledger entries, and identify discrepancies, substituting human auditors' manual work with intelligent automation that maintains accuracy while dramatically reducing time
Solution Approach 2:
The system enables self-service verification by automatically retrieving and validating its own data sources. The AI models independently verify data integrity without requiring manual intervention, and the system can autonomously identify and flag discrepancies for auditor review, reducing the need for time-consuming manual verification
Data Source
AI summary
A method is disclosed for analysing a data set to determine a first processes. First messages are provided, the first messages classified into a plurality of different classes with a plurality of different likelihoods, a single first message classified into different classes based on different criteria. From the first messages a first subset of the first messages is retrieved based on a combination of one or more classifications, a likelihood of the one or more classifications, and another classification for messages within the first subset of the first messages. The likelihood of the classifications has more than two (2) potential values.