Probabilistic Customer Activity Classification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Measuring the effectiveness of automated customer support systems is challenging due to the reliance on customer surveys, which can be biased and do not accurately reflect user interactions, especially in determining success or failure in solving issues.

Innovation Solution

Classifying customer activity in automated systems using probabilistic models, such as Hidden Markov Models, that analyze observable measurements like document interactions and affective states, allowing for unbiased classification of user success or failure without user intervention.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If customer surveys or interviews are used to measure automated customer support systems, then customer feedback can be obtained, but the measurements become biased and do not accurately reflect user interactions

Engineering Contradiction:

Improveaccuracy of success measurementVSAvoidbias in customer feedback

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The system automatically classifies customer contacts using probabilistic models that analyze interaction traces without requiring customer participation in surveys. The automated system serves itself by measuring its own performance through objective analysis of customer behavior patterns, eliminating the need for biased customer feedback.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical survey-based measurement system with an automated probabilistic classification system. Instead of relying on customers to manually provide feedback through surveys, the system uses computational models to automatically analyze interaction traces and determine success metrics.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If automated classification systems are implemented, then measurement objectivity is improved, but the complexity of analyzing interaction traces increases

Engineering Contradiction:

Improveobjectivity of classificationVSAvoidcomplexity of probabilistic models

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The probabilistic classification system is designed to handle multiple types of interaction traces (documents read, menu options chosen, information provided) through a unified modeling approach. The same probabilistic framework can classify different contact types and analyze various interaction patterns, reducing the need for separate specialized systems.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system transforms complex interaction trace data into simplified probabilistic parameters that can be efficiently processed. By converting detailed interaction sequences into probability distributions and classification scores, the system manages complexity while maintaining measurement precision.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS8938405B2Classifying activity using probabilistic models

Publication Date: 2015.01.20 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

A method, an apparatus and an article of manufacture for classifying customer activity in an automated customer support system. The method includes obtaining input from the automated customer support system, wherein the input comprises an observable measurement of customer activity in the automated customer support system, computing a probability that the input corresponds to one of one or more probabilistic models, and using the computed probability to classify the customer activity in the automated customer support system by considering the probabilistic model corresponding to a highest computed probability.