ML Journal Approval Control for Faster Financial Closing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic journal systems rely heavily on human review and hardcoded programmed rules to approve or reject journal entries, which is time-consuming and prone to errors, requiring significant manual effort and reprogramming for rule changes.
Innovation Solution
A machine learning model is trained on historical journal entries to predict whether entries can be automatically approved, using feature vectors and pattern recognition to determine eligibility for automatic approval, reducing the need for manual review and hardcoded rules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If human review and hardcoded programmed rules are used to approve journal entries, then reliability of approval decisions is maintained, but processing time and resource consumption increase significantly
Solution Approach 1:
A machine learning model is introduced as an intermediary between journal entries and human reviewers. The model predicts approval likelihood and generates explanations, filtering entries that require human review. This intermediary processes entries automatically while maintaining reliability through confidence thresholds and explainable AI techniques.
Solution Approach 2:
The system enables journal entries to be approved automatically without human intervention when the ML model predicts high confidence. The model serves itself by learning from historical data and making autonomous approval decisions, reducing the need for manual review of routine entries.
2Stability of the object's composition
If hardcoded programmed rules are used for journal entry approval, then consistency in decision-making is maintained, but adaptability to new patterns decreases
Solution Approach 1:
The system transitions from static hardcoded rules to a dynamic machine learning model that continuously learns from new data. The model adapts to changing patterns in journal entries while maintaining decision consistency through its learned decision boundaries and confidence scoring mechanism.
Solution Approach 2:
Instead of changing programmed rules, the system changes the parameters of the ML model through continuous training on historical data. The model's internal parameters (weights and biases) are automatically adjusted to reflect new patterns, providing adaptability without requiring manual rule modifications.
3Measurement precision
If manual review of each journal entry is performed, then accuracy of approval decisions is improved, but productivity of the system decreases
Solution Approach 1:
The system applies partial automation by using the ML model to pre-screen and score all entries, then directing human reviewers only to entries with low confidence scores or high risk indicators. This partial application of automated review maintains accuracy for critical entries while improving overall productivity.
Solution Approach 2:
The system implements feedback loops where human reviewer decisions on ML-predicted entries are fed back into the training data. This continuous feedback improves the model's accuracy over time, allowing progressively higher automation levels while maintaining or improving approval accuracy.
4Reliability
If experienced programmers create and maintain programmed rules, then reliability of rule implementation is ensured, but device complexity and maintenance burden increase
Solution Approach 1:
The system replaces the mechanical process of programming and maintaining explicit rules with a machine learning model that learns patterns automatically from data. This substitution eliminates the need for programmers to manually create and update complex rule sets, reducing system complexity while maintaining reliability through data-driven decision-making.
Data Source
AI summary
Systems, methods, and other embodiments associated with predicting, controlling, and optimizing journal closing are described. In one embodiment, a method includes training a machine learning model based on historical journal entries, wherein the machine learning model learns combinations of features from journal entries that were approved and journal entries that were not approved. The model may evaluate a target set of journal entries using extracted features from individual journal entries. The model predicts which of the individual journal entries qualify for automatic approval by identifying combinations of the extracted features from the individual journal entries that have a multi-dimensional similarity to the learned combinations of features that were previously approved from the historical journal entries. Entries that are automatically approved may be marked with an automatically approved status and may be automatically transferred to a closing process.