ML Date Classification for Tax Effective Date Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying tax effective dates in voluminous tax laws and regulations are inefficient, often requiring manual reading and prone to errors, while keyword searches are inaccurate.
Innovation Solution
A computerized system using trained machine learning models to extract and classify candidate dates, including a first model for date extraction and a second model for classification, and a regular expression engine to identify tax rates and amounts, creating a modified article with visually distinguishable highlights.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual reading of tax laws and regulations is performed, then accuracy in identifying tax rates and effective dates is improved, but time consumption and cost increase significantly
Solution Approach 1:
The patent introduces an intermediary system consisting of trained machine learning models that act as a mediator between the tax law text and the tax expert. The first ML model extracts candidate dates, the second ML model classifies them as tax effective dates or other dates, and the regular expression engine extracts tax rates. This intermediary automated system performs the initial identification work with high accuracy, allowing tax experts to verify results rather than manually reading entire documents, thus resolving the contradiction between accuracy and time consumption.
Solution Approach 2:
The system performs preliminary action by automatically extracting and classifying candidate dates and tax rates before presenting them to tax experts for verification. The ML models pre-process the voluminous text, identify potential tax effective dates and rates, and organize them in a structured format. This preliminary automated extraction reduces the burden on tax experts, who only need to review and verify the pre-identified candidates rather than manually searching through entire tax law documents.
2Productivity
If keyword searching is used to identify tax rates and effective dates, then time consumption is reduced, but accuracy decreases due to missing or misidentifying certain rates and dates
Solution Approach 1:
The patent replaces the mechanical keyword searching system with an intelligent machine learning-based system. Instead of relying on simple text matching algorithms that can miss context-dependent information, the system uses trained ML models that understand the semantic meaning and context of dates and rates within tax law documents. The first ML model (e.g., spaCy NER) extracts candidate dates based on contextual understanding, and the second ML model classifies them by analyzing surrounding text patterns, thereby maintaining high identification speed while significantly improving accuracy over traditional keyword searching.
3Reliability
If manual reading is performed to reduce errors, then reliability is improved, but productivity decreases due to significant time consumption
Solution Approach 1:
The system implements self-service by enabling the automated extraction and classification of tax effective dates and rates without requiring manual reading of entire tax law documents. The trained ML models independently process the text, identify candidate dates, classify them as tax effective dates or other dates, and extract tax rates. This self-service capability maintains high reliability by consistently applying the same classification logic, while dramatically improving productivity by processing documents much faster than manual reading.
4Productivity
If automated systems are used to process tax laws, then productivity is improved, but complexity of the system increases
Solution Approach 1:
The patent applies segmentation by dividing the automated processing system into three distinct, modular components: (1) the first ML model for extracting candidate dates, (2) the second ML model for classifying dates as tax effective or other dates, and (3) the regular expression engine for extracting tax rates. Each component has a specific, well-defined function and can be independently trained, tested, and maintained. This segmentation reduces system complexity by making each component manageable and interchangeable, while still achieving high productivity through their coordinated operation.
Data Source
AI summary
A computing system is provided, including a processor configured to, at inference time, input an article to a first trained machine learning (ML) model. The processor is further configured to extract, via the first trained ML model, candidate dates from the article. The processor is further configured to input the extracted candidate dates to a second trained ML model, and classify, via the second trained ML model, each extracted candidate date into one of plurality of classifications including a tax effective date and a date other than a tax effective date. The processor is further configured to create a modified article that highlights the candidate date classified as the tax effective date and the candidate date classified as the date other than the tax effective date, in which the tax effective date and the date other than the tax effective date are highlighted in a visually distinguishable manner.