Automated Call Transcription Data Extraction Using ML and Sentiment Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Call center agents manually identify personal information and key words from transcriptions, reducing productivity and instant recall in subsequent calls due to the lack of sentient analysis and AI/ML utilization.

Innovation Solution

A computer-based system using natural language processing, tonal rule engines, sentiment analysis, and TF-IDF algorithms to automatically extract personal information, determine key terms, and generate call scripts for subsequent interactions, reducing manual input and improving efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If agents manually identify personal information and key words from transcriptions, then accuracy of information extraction is maintained, but productivity of the call center decreases

Engineering Contradiction:

Improvecall center productivityVSAvoidmanual information extraction

Core Design Contradiction:

ProductivityVSExtent of automation

Solution Approach 1:

The system enables self-service automated extraction of personal information and key terms from call transcriptions using machine learning models and natural language processing, eliminating the need for manual agent intervention while maintaining high accuracy in information identification

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical manual process of information extraction with an automated computational system using trained machine learning models, tonal rule engines, and sentiment analysis algorithms to identify and extract personal information and key terms from transcription data

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Loss of information

If manual identification of personal information is performed, then instant recall of information is reduced, but the process requires minimal technological infrastructure

Engineering Contradiction:

Improveinstant recall of key informationVSAvoidAI/ML system complexity

Core Design Contradiction:

Loss of informationVSDevice complexity

Solution Approach 1:

The system performs preliminary automated extraction and structuring of personal information and key terms from call transcriptions before subsequent calls occur, creating ready-to-use structured data that enables instant recall and improves the quality of customer interactions in future calls

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary automated processing layer between the call transcription and the agent, using machine learning models and natural language processing to extract and structure information, thereby preventing information loss and enabling instant recall without requiring the agent to manually process the data

Inventive Principle:

Principle #24Intermediary (Mediator)

3Productivity

If automated extraction using AI/ML is implemented, then productivity is enhanced, but the system complexity increases

Engineering Contradiction:

Improveinformation extraction efficiencyVSAvoidnatural language processing system

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the automated information extraction system into distinct functional modules including a trained machine learning model for personal information identification, a tonal rule engine for key term determination, and sentiment analysis components, allowing each segment to specialize in specific extraction tasks and improving overall system efficiency

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system employs a multi-functional processing architecture where the trained machine learning model handles personal information extraction, the tonal rule engine identifies key terms, and sentiment analysis provides additional context, with all components working together to comprehensively process transcription data and generate structured output

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20250023983A1Computer-based systems configured for automated extraction of data from inputs and methods of use thereof

Publication Date: 2025.01.16 CAPITAL ONE SERVICES LLC

AI summary

In some embodiments, the present disclosure provides an exemplary method that may include steps of receiving an input text data retrieved from a transcription associated with a previously recorded audio data file between a user of a plurality of users and an agent associated with a call center; identifying personal information associated with the user of the plurality of users from the input text data by inputting the input text data into a trained machine learning model; determining at least one key term within the personal information associated with the user of the plurality of users; automatically determining a confidence positivity score associated with the at least one key term; automatically extracting a plurality of tuples from the input text data; storing the plurality of tuples in an external database; and automatically generating a call script for conducting a subsequent call with the user.