AI-ML Data Flow Prediction for Privacy Policy Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The data sharing economy is complex and opaque, leading to consumers' personally identifiable information and behavioral data being shared with unknown third parties without their knowledge or consent, resulting in a cascading effect of data reselling without explicit consent.
Innovation Solution
A system and method using AI-ML to track and manage data sharing by receiving customer input on first-level entities, processing privacy policies and data sources, generating a data graph, predicting data flow, and providing insights and recommendations to consumers on data sharing relationships and usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If vendors share consumer data with third parties to expand the data sharing economy, then the quantity of data available for commercial use increases, but consumer privacy protection deteriorates as data is shared with unknown parties without explicit consent
Solution Approach 1:
The patent introduces an intermediary system that sits between consumers and the data sharing network. This intermediary uses AI-ML models to analyze privacy policies, predict data flows, and provide consumers with actionable insights about where their data may end up, thereby mediating the information asymmetry without blocking the data economy itself
Solution Approach 2:
The system implements feedback by continuously monitoring data sharing practices through privacy policy analysis and using AI-ML models to predict data flows. This feedback loop provides consumers with real-time information about their data's journey through the sharing economy, enabling informed decisions about data sharing preferences
2Adaptability or versatility
If the data sharing network expands to include more third parties and resellers, then the versatility of data usage increases, but the complexity of tracking data flows increases
Solution Approach 1:
The patent replaces manual tracking mechanisms with AI-ML-based automated analysis systems. These systems use natural language processing to parse privacy policies and machine learning models to predict data flows, substituting complex manual tracking efforts with intelligent automated systems that can handle the expanding data sharing network
Solution Approach 2:
The AI-ML framework serves multiple functions: it analyzes privacy policies, predicts data flows, identifies potential unauthorized sharing, and provides consumer insights. This multi-functional system handles the complexity of tracking across diverse data sharing relationships without requiring separate mechanisms for each function
3Loss of information
If consumers are provided with detailed information about data sharing relationships, then transparency improves, but the difficulty of processing and understanding policy information increases
Solution Approach 1:
The system enables consumers to self-serve by providing them with AI-generated insights about their own data sharing relationships. Rather than requiring consumers to manually analyze complex privacy policies, the system automatically processes this information and presents it in an understandable format, allowing consumers to make informed decisions without becoming experts in policy analysis
Data Source
AI summary
Systems, methods, and devices for tracking and managing data shared with third parties are disclosed. In one embodiment, a method including: retrieving data collection and usage policies of an entity; processing the data collection and usage policies with a natural language processing (NLP) model; generating, by the NLP model, predictive data collection and data usage attributes; generating a feature vector from the predictive data collection and data usage attributes; processing the feature vector with a graph neural network; storing data structured as a graph including the entity and the predictive data collection and data usage attributes; and processing the data structured as a graph with a classifier model that labels the entity as a first node in the data structured as a graph and predicts an edge to a second node in the data structured as a graph based on the predictive data collection and data usage attributes.