Account Relationship Matrix for Group Cheating Detection

The patent merges multiple account analyses into a unified framework by constructing an account relationship matrix that integrates identity-related information from multiple accounts. This allows the system to detect group cheating patterns by analyzing relationships between accounts rather than treating each account independently, thereby improving anti-cheating effectiveness against organized cheating gangs.

The patent transitions from single-account analysis to multi-account relationship analysis by introducing an account relationship matrix as an additional dimension. This matrix captures relationships between accounts based on identity features, enabling the system to detect cheating patterns that span multiple accounts and identify key cheating accounts within groups.

The patent segments the complex multi-account analysis into distinct computational steps: extracting identity-related information from historical upload data, constructing the account relationship matrix based on identity features, calculating importance degrees for each account, and identifying key cheating accounts. This segmentation reduces processing complexity by breaking down the overall task into manageable stages with clear inputs and outputs.

The patent applies local quality by focusing computational resources on identifying key cheating accounts within the group rather than treating all accounts equally. By calculating importance degrees for each account based on their relationships and behaviors, the system prioritizes analysis of accounts that are most likely to be involved in cheating, thereby improving identification accuracy while reducing overall computational burden.