Unstructured Text Complaint Classification via Hybrid Scoring

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for analyzing unstructured computer text are inefficient and inaccurate in identifying and classifying complaint-specific interactions, as they typically only review text at a high level without advanced semantic analysis.

Innovation Solution

A system utilizing a combination of rules-based and machine learning techniques to analyze unstructured text, including tokenization, sentiment scoring, and word vector analysis in a high-dimensional space, to identify and classify complaint-specific interactions, tonality, and respondent profiling.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual interpretation of unstructured text is used, then accuracy in understanding context and sentiment is improved, but productivity and efficiency deteriorate

Engineering Contradiction:

Improveaccuracy in understanding context and sentimentVSAvoidefficiency in processing text

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The patent introduces an intermediary system comprising NLP processing, machine learning models, and sentiment analysis algorithms that act as a mediator between raw unstructured text and actionable insights. This intermediary automatically processes text to extract context, themes, and sentiment while maintaining high accuracy through advanced computational techniques, thereby resolving the contradiction between manual accuracy and automated productivity.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Productivity

If computerized text analysis techniques are used, then productivity is improved, but measurement precision and accuracy deteriorate

Engineering Contradiction:

Improveautomation of text analysisVSAvoidaccuracy in classifying complaints

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies parameter changes by continuously refining multiple technical parameters including training datasets, model architectures, sentiment scoring thresholds, and classification criteria. By optimizing these parameters through iterative machine learning processes and feedback loops, the system achieves both high productivity through automation and high measurement precision in complaint classification, effectively resolving the contradiction between automated processing and accurate classification.

Inventive Principle:

Principle #35Parameter changes

3Device complexity

If basic complaint type classification is used, then device complexity is reduced, but adaptability and versatility deteriorate

Engineering Contradiction:

Improvesimplicity of classification systemVSAvoidability to classify specific complaint types

Core Design Contradiction:

Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent implements segmentation by dividing the complaint classification system into multiple hierarchical levels: broad complaint categories, specific complaint types, and detailed sub-classifications. This segmented approach allows the system to maintain manageable complexity at each level while achieving high adaptability and versatility through the cumulative effect of multiple classification dimensions, effectively resolving the contradiction between system simplicity and classification capability.

Inventive Principle:

Principle #1Segmentation

4Measurement precision

If advanced NLP and machine learning techniques are applied, then measurement precision in text analysis is improved, but device complexity increases

Engineering Contradiction:

Improveaccuracy in text interpretationVSAvoidcomplexity of processing system

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies dynamics by implementing adaptive machine learning models that dynamically adjust their processing complexity based on the characteristics of the input text and the specific analysis task. The system dynamically selects and applies appropriate NLP techniques, adjusts model parameters in real-time, and optimizes computational resources according to the complexity of the text being analyzed, thereby achieving high measurement precision while managing device complexity through dynamic adaptation.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS10692016B2Classifying unstructured computer text for complaint-specific interactions using rules-based and machine learning modeling

Publication Date: 2020.06.23 FMR CORP

AI summary

Methods and apparatuses are described for analyzing unstructured computer text for identification and classification of complaint-specific interactions. A computer data stores unstructured text. A server computing device splits the unstructured text into phrases of words. The server generates a set of tokens from each phrase and removes tokens that are stopwords. The server generates a normalized sentiment score for each set of tokens. The server uses a rules-based classification engine to generate a rules-based complaint score for each set of tokens. The server uses an artificial intelligence machine learning model to generate a model-based complaint score for each set of tokens. The server determines determine whether each set of tokens corresponds to a complaint-specific interaction based upon the rules-based complaint score and the model-based complaint score.