Aggregate Rating Filtering via Fraud Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for aggregating ratings from different review platforms do not provide an optimal aggregate score due to variations in reviewers' approaches and reliability, leading to inaccurate credibility assessments of entities like physicians.
Innovation Solution
A computer-implemented method and system that retrieves and filters rating indicators from various repositories, including web-based and social networking platforms, using a special purpose processor to identify and exclude fraudulent indicators based on parameters like geo-location, gender, frequency, and sentiment writing style, and then aggregates the remaining indicators to generate an accurate aggregate rating.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all ratings from different review platforms are aggregated using simple averaging or summation, then the aggregation process is simple and fast, but the accuracy of the aggregate score deteriorates due to reviewer reliability variations and fraudulent ratings
Solution Approach 1:
The patent segments the aggregation process into distinct functional modules: a filtering engine that segments fraudulent ratings from legitimate ones using multiple parameters (reviewer reliability scores, rating patterns, temporal analysis), and a separate aggregation engine that processes only filtered ratings. This segmentation allows complex fraud detection to be isolated from the aggregation process, improving accuracy without entirely overwhelming the system.
Solution Approach 2:
The patent introduces an intermediary filtering layer between rating collection and aggregate score generation. This filtering engine acts as a mediator that processes raw ratings through multiple validation stages (reviewer authentication, pattern recognition, reliability scoring) before passing clean data to the aggregation function, thereby improving accuracy while managing complexity through modular design.
2Reliability
If multiple parameters are used to filter fraudulent ratings, then the reliability of the aggregate rating improves, but the computational complexity and processing time increase
Solution Approach 1:
The patent implements preliminary action by pre-computing and storing reviewer reliability scores, historical rating patterns, and authentication data before the actual aggregation process. When new ratings arrive, the system quickly compares them against pre-established criteria and trusted reviewer lists, enabling fast filtering without real-time complex computations, thus reducing processing time while maintaining high reliability.
3Measurement precision
If fraudulent rating indicators are not filtered out, then the aggregation process remains simple, but the credibility assessment of entities becomes inaccurate
Solution Approach 1:
The patent applies local quality by implementing different filtering strategies for different types of ratings and reviewers. Instead of a single uniform complex filter, the system uses tailored validation rules: authenticated reviewers get lighter validation, while anonymous reviewers undergo stricter pattern matching. Geographic location analysis and entity-specific trust weights are applied locally to relevant ratings, improving credibility assessment without requiring universally complex processing for all ratings.
Data Source
AI summary
A computer-implemented method and system for publishing an aggregate rating indicator based on a plurality of discrete rating indicators associated with a rated entity. The method includes searching a plurality of repositories containing the plurality of discrete rating indicators associated with the rated entity. The method may include retrieving a first rating indicator, a second rating indicator, and a third rating indicator. At least one of the first rating indicator, the second rating indicator, and the third rating indicator may be a fraudulent rating indicator. The method may include filtering out the fraudulent rating indicator from among the first, second, and third rating indicators, wherein the fraudulent rating indicator is determined based on a combination of parameters. The method may include aggregating the first, second, and third rating indicators after filtering out the fraudulent rating indicator such that the aggregation results in an aggregate rating indicator.