Verbal Feedback Classification Using Artificial-Text Transfer Learning

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing customer feedback systems require manual triage to filter out non-actionable 'noise' from verbal feedback, consuming significant time and resources, especially in high-volume scenarios.

Innovation Solution

A machine learning model that classifies free-text feedback using a combination of free-text and artificial-text features, generated from parameter values, to automate the triage process, reducing dimensionality and improving classification accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual triage is used to filter noise from verbal feedback, then classification accuracy is improved, but processing time and resource requirements increase significantly

Engineering Contradiction:

Improveclassification accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the feedback processing task into two distinct components: free-text features from user feedback and artificial-text features generated from parameter values. This segmentation allows the system to process structured parameter data separately from unstructured text, enabling more efficient automated classification while maintaining accuracy through combined feature analysis.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces artificial-text as an intermediary element that bridges structured parameter values and the classification task. By generating artificial-text representations from parameter values using template-based methods, the system creates additional informative features that assist the machine learning model in achieving accurate classification without requiring manual triage of all feedback.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If manual triage is used to filter noise from verbal feedback, then classification quality is improved, but resource requirements increase significantly

Engineering Contradiction:

Improveclassification qualityVSAvoidresource requirements

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system implements self-service by automatically generating artificial-text features from parameter values without human intervention. The template-based generation process and machine learning classification operate autonomously, eliminating the need for manual triage resources while maintaining classification quality through the combined feature approach.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical manual triage process with an automated computational system. Instead of human specialists manually reviewing and filtering feedback, the system uses machine learning models processed by computers to perform classification, significantly reducing resource requirements while maintaining or improving classification quality.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Ease of operation

If only free-text features are used for classification, then processing simplicity is maintained, but classification accuracy decreases

Engineering Contradiction:

Improveprocessing simplicityVSAvoidclassification accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The patent merges free-text features and artificial-text features into a unified feature set for classification. By combining these two complementary feature types, the system achieves improved classification accuracy while maintaining processing simplicity through automated feature generation and integrated model training that handles both feature sources uniformly.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS20260094061A1Identifying noise in verbal feedback using artificial text from non-textual parameters and transfer learning

Publication Date: 2026.04.02 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

Methods and systems are provided for classifying free-text content using machine learning. Free-text content (e.g., customer feedback) and parameter values organized according to a schema are received. A free-text corpus is generated, and an artificial-text corpus is generated by applying rules to the parameter values. The artificial-text corpus is generated by converting the parameter values into a finite set of words based on the rules and concatenating the words of the finite set of words into a fixed sequence wordlist. Feature vectors (e.g., sentence embeddings) based on the free-text corpus and the artificial-text corpus are combined and forwarded to a machine learning model for classification. The machine learning model may be trained with a bias towards a specified metric (e.g., precision, recall, F1 score). The model may be trained using transfer learning with training data from a different category of free-text content (e.g., a different category of customer feedback).