Trust Flow Categorization System for Information Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for categorizing information are either computationally expensive or overly simplistic and prone to error, failing to effectively filter out uninteresting or malicious information items such as spam and phishing attempts.
Innovation Solution
A system utilizing a trust flow module to assign and propagate trust values across interlinked information items, with a normalization module to refine these values, allowing for accurate categorization and filtering of information based on accumulated trust values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current categorization methods are used, then information can be categorized, but the methods are either computationally expensive or overly simplistic and prone to error
Solution Approach 1:
The patent introduces trust values as an intermediary metric that mediates between the complexity of analysis and the accuracy of categorization. Instead of directly categorizing information items through complex computational methods, the system assigns trust values based on source reliability and propagates these values through the information network, enabling accurate categorization with reduced computational overhead.
Solution Approach 2:
The system changes the parameter of categorization from direct content analysis to trust value propagation. By transforming the categorization problem into a trust value assignment and propagation problem, the system achieves more reliable categorization results while avoiding the computational expense of traditional methods.
2Object-affected harmful factors
If spam filters are used to remove harmful information, then the volume of information is reduced, but the filters may incorrectly categorize legitimate information as malicious
Solution Approach 1:
The system performs preliminary trust value assignment to information sources before filtering occurs. By pre-establishing trust values for sources and propagating them to their content, the system creates a foundation for more accurate filtering decisions, reducing false positives while maintaining effective malicious information filtering.
Solution Approach 2:
The trust value propagation system incorporates feedback mechanisms where categorization results and filter performance inform ongoing trust value adjustments. This feedback loop allows the system to learn from past filtering decisions and improve accuracy over time, reducing incorrect categorization of legitimate information.
3Measurement precision
If trust values are propagated iteratively through information items, then categorization accuracy improves, but the computational time increases
Solution Approach 1:
The system applies partial iteration by performing trust value propagation for a limited number of iterations or until convergence threshold is reached. This partial action approach achieves sufficient categorization precision without the excessive computational time required for complete or exhaustive iteration through the entire information network.
Data Source
AI summary
A system for the categorization of interlinked information items, the system comprising: a trust flow module which is configured to receive a seed trust list of one or more first information items, the seed trust list associating the one or more first information items with one or more categories; and a trust flow module configured to: associate a respective trust value with each of the one or more categories for the one or more first information items; and iteratively pass at least part of the or each trust value to one or more further information items to generate, for each of the one or more further information items, at least one accumulated trust value associated with a category of the one or more categories, such that the one or more further information items can be categorized based on the at least one accumulated trust value and associated category.