Spam Detection via Aggregated Blocked Lists
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech recognition systems are ineffective in identifying and mitigating spam communications, as users' blocking of spam numbers does not share across devices and platforms, allowing spammers to continue sending unsolicited calls and messages by disguising their origins or using automated bots.
Innovation Solution
A system that aggregates blocked communications lists from multiple users to identify and flag spam numbers based on frequency of blocking, and uses audio analysis to detect overlapping communications patterns indicative of spam activity, thereby throttling or disabling suspected spammer IDs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If users block spam numbers individually on their own devices, then each user can protect themselves from known spam numbers, but spam communications continue to affect other users and spammers can easily evade blocking by using different numbers or bots
Solution Approach 1:
The patent combines isolated user-level blocking actions into a centralized cloud-based blocking system. When one user blocks a spam number, that information is aggregated and shared across the entire user base through the cloud platform, transforming individual protective actions into collective spam mitigation without requiring complex changes to user devices
Solution Approach 2:
The cloud-based platform serves as an intermediary between users and spam numbers. Instead of users directly managing blocking on their devices, the cloud platform mediates by collecting blocking data, analyzing patterns, and distributing updated blocking lists to all users, thereby simplifying the system architecture while improving reliability
2Measurement precision
If speech recognition systems process all incoming communications to identify spam, then spam detection capability improves, but system processing time and computational resources increase significantly
Solution Approach 1:
The system performs preliminary spam detection by analyzing communication patterns, caller IDs, and messaging behaviors before fully processing the communication content. This preliminary filtering uses lightweight algorithms to identify likely spam, allowing the system to skip intensive speech recognition processing for suspicious communications, thereby maintaining high detection accuracy while reducing processing time
Solution Approach 2:
The spam detection process is segmented into multiple stages: initial pattern matching, intermediate analysis, and detailed speech recognition only for suspicious cases. This segmentation allows the system to quickly filter out obvious spam using simple criteria while applying more resource-intensive processing only when necessary, optimizing the balance between accuracy and processing time
Data Source
AI summary
Techniques for detecting spam accounts in a system are described. When a system creates a user profile, the system may ingest a blocked communications list. The system may determine how many times each blocked communications number represented in the ingested blocked communications list is included in blocked communications lists of various users of the system. If a blocked communications number represented in the ingested blocked communications list is included in at least a threshold number of other blocked communications lists, the system may mark the communications number as spam at a system level and engage in appropriate mitigation techniques (e.g., throttle the phone numbers activity, disable the phone number's ability to communicate with system devices, etc.).