Ethics Violation Detection via NLP and Machine Learning Prioritization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for evaluating ethics violations within organizations are inefficient, subjective, and inconsistent due to disparate reporting systems, lacking a standardized approach for claim evaluation and prioritization.
Innovation Solution
A system that preprocesses claims from various sources into a common digital format, utilizing natural language processing and machine learning models, including binary, multi-class, and multi-label models, to detect and prioritize ethics violations, providing a standardized interface for investigation and visualization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple disparate reporting systems are used to collect ethics claims, then the system can receive claims from various sources, but the evaluation becomes inefficient and inconsistent
Solution Approach 1:
The patent introduces a centralized ethics management platform that acts as an intermediary between disparate reporting systems and the evaluation process. This platform standardizes claim intake from multiple sources (hotlines, emails, forms) into a unified format, enabling efficient processing while maintaining versatility in receiving claims from various channels.
Solution Approach 2:
The ethics management platform performs multiple functions within a single system: it receives claims from diverse sources, preprocesses and standardizes data, applies machine learning models for detection and prioritization, and generates visualizations. This multi-functional approach improves evaluation efficiency while maintaining adaptability to different reporting formats.
2Measurement precision
If manual evaluation of ethics claims is performed, then subjective assessment can be applied, but the process becomes time-consuming and inconsistent
Solution Approach 1:
The patent replaces manual mechanical evaluation with automated machine learning models. Binary classification models determine whether claims represent ethics violations, while multi-class models prioritize cases. This substitution eliminates subjective variability and significantly reduces evaluation time while maintaining consistent application of ethics standards.
Solution Approach 2:
The system enables self-service evaluation through automated processing. The machine learning models independently assess and prioritize claims without requiring manual review for every case, allowing the system to handle high volumes of claims efficiently and consistently.
3Reliability
If comprehensive analysis of all claims is performed, then thorough evaluation is achieved, but resource allocation becomes inefficient
Solution Approach 1:
The patent applies different levels of analysis to different claims based on their characteristics. Machine learning models identify high-priority cases requiring thorough manual review while automatically processing lower-priority claims. This localized quality approach ensures comprehensive evaluation of critical cases while efficiently handling routine matters, optimizing resource allocation.
Solution Approach 2:
The system performs partial automated analysis on all claims through machine learning models, then applies excessive thoroughness only to high-priority cases identified by the models. This tiered approach maintains reliability for critical cases while improving overall productivity by avoiding exhaustive analysis of every claim.
4Measurement precision
If standardized formatting is applied to all claims, then consistent evaluation is enabled, but data preprocessing complexity increases
Solution Approach 1:
The patent introduces a preprocessing layer that acts as an intermediary between diverse input formats and the evaluation models. This layer automatically standardizes claims from different sources into a unified format, enabling consistent evaluation while managing preprocessing complexity through automated transformation rules and natural language processing techniques.
Data Source
AI summary
An system and a method for the detection and visualization of reported ethics cases is disclosed. The system receives a set of digital records corresponding to a reported ethics violations. The system converts each of the digital records from the set of digital records into a common digital format. The system deconstructs the uniform text structure of each digital recorded by a natural language processing module to lemmatize words, remove punctuation, and remove stop words. The system inputs each deconstructed uniform text structure into a binary machine learning data model. The system inputs each deconstructed uniform text structure into a multiclass machine learning data model. The system inputs the determined value and the label to an ensemble machine learning data model. The system prioritizes reported ethics violations into one or more lists based on the determination of the possible class and transmits the list to a user interface.