Caller Trust Scoring via NLP Voice Pattern Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solutions fail to effectively identify and filter out spam calls, particularly when callers make fraudulent statements, as they lack the capability to analyze and compare the content of calls for trustworthiness.
Innovation Solution
A system that utilizes natural language processing (NLP) to analyze the voice patterns and topics of conversations, assigning a trust score to callers based on their previous statements, using a Hilbert space transformation and Dirichlet distribution to determine the trustworthiness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing call filtering solutions are used, then calls are blocked based on simple criteria, but they fail to identify spam calls when callers make fraudulent statements
Solution Approach 1:
The patent introduces an NLP processing system as an intermediary between the raw call data and the trust score calculation. This intermediary layer analyzes call transcripts, extracts topics, and compares them against known spam patterns, enabling reliable spam identification without requiring complex manual analysis of each call
Solution Approach 2:
The patent replaces simple mechanical filtering rules with intelligent NLP-based analysis. Instead of using basic keyword matching or blacklist/whitelist approaches, the system uses natural language processing to understand call content, identify fraudulent statements, and calculate trust scores, significantly improving reliability
2Measurement precision
If NLP analysis is applied to analyze call content, then trust scores can be calculated, but processing time increases
Solution Approach 1:
The patent segments the NLP analysis process into distinct stages: call transcript extraction, topic identification, comparison against known patterns, and trust score calculation. This segmentation allows each stage to be optimized independently and enables parallel processing of multiple calls, reducing overall processing time while maintaining precision
Solution Approach 2:
The system performs preliminary analysis by pre-processing call transcripts and identifying key topics before the full trust score calculation. By preparing data in advance and organizing it into structured formats, the system reduces the computational burden during the actual scoring phase, minimizing processing time
Data Source
AI summary
A computer captures a voice of a user. The computer determines a frequency spectrum and a voice pattern of the voice. The computer identifies one or more topics of the voice by transcribing the voice by a natural language processing. The computer identifies the user based on matching the frequency spectrum of the voice to the frequency spectrum of the conversation and the pattern of the voice to the pattern of the conversation when a conversation is intercepted and determines a trust score based on comparing the one or more topics to the one or more topics extracted from the conversation.