Voice Call Complexity Scoring Through Topic Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional contact center performance metrics lack a measure for call complexity, which affects customer satisfaction and agent performance, as they cannot accurately quantify the complexity of interactions based on the number and complexity of topics discussed during calls.
Innovation Solution
A CRM system processes call transcripts to identify topics and transitions, calculating a complexity score by analyzing topic changes and revisits, providing a decimal value that represents the interaction's complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If conventional performance metrics (AHT, form completion time) are used to measure agent performance, then ease of calculation and implementation is improved, but measurement precision of call complexity is worsened
Solution Approach 1:
The patent replaces manual or simple metric-based complexity assessment with automated natural language processing and machine learning algorithms that analyze call transcripts, speaker turns, and topic transitions to compute complexity scores, thereby achieving precise measurement without manual intervention
Solution Approach 2:
The patent introduces intermediate metrics such as topic transition frequency, speaker turn length, and information density as mediators between the raw call data and the final complexity measurement, enabling accurate complexity assessment through multiple layered analyses
2Measurement precision
If detailed transcript analysis is performed to measure call complexity accurately, then measurement precision is improved, but loss of time for processing is worsened
Solution Approach 1:
The patent performs preliminary processing of call transcripts by pre-identifying key features such as topic transitions, speaker turns, and semantic segments before the main complexity calculation, thereby reducing the computational burden and processing time of the subsequent analysis
Solution Approach 2:
The patent divides the call transcript into discrete segments such as topic clusters, speaker turns, and information units, allowing parallel processing and efficient computation of complexity metrics across multiple segments simultaneously
3Measurement precision
If multiple topics and topic changes are tracked to compute complexity score, then measurement precision of call complexity is improved, but device complexity of the system is worsened
Solution Approach 1:
The patent implements a universal topic modeling framework that can identify and track multiple topics, their transitions, and their relationships using a single set of NLP algorithms and processing pipelines, thereby achieving precise complexity measurement without proportionally increasing system complexity
Solution Approach 2:
The patent uses parameter-based topic modeling where topics are represented as vectors of parameters or keywords, allowing efficient comparison and tracking of topic transitions through parameter similarity calculations rather than complex semantic analysis
Data Source
AI summary
An event prioritization system and methods are provided that are configured to dynamically compute call complexity for voice data calls using topics identified from call transcripts. The system includes a processor and a computer readable medium operably coupled thereto, the computer readable medium comprising a plurality of instructions stored in association therewith that are accessible to, and executable by, the processor, to perform call evaluation operations which include receiving a transcript of a voice data call, parsing the transcript for at least one topic, generating, based on the parsing, a list of topics, analyzing, by a call complexity evaluation engine, the list of topics for a complexity rating, determining, based on the complexity rating and a length of the voice data call, a deviation from an expectation, flagging the voice data call, and outputting the flagged voice data call via an interface.