Aspect Term Sentiment Analysis Using Multi-Model Attention Weights

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current information processing systems face challenges in efficiently analyzing and managing unstructured text data, particularly in identifying sentiment and aspect terms, due to the need for manual screening and rule-based systems, which are tedious and time-consuming for large volumes of data.

Innovation Solution

The system employs a multi-machine learning model approach to perform sentiment analysis by encoding unstructured text data, classifying words as aspect terms or non-aspect terms, determining attention weights for surrounding words, and generating sentiment classifications based on these encodings, enabling automated and efficient sentiment analysis for aspect terms.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If manual screening and rule-based systems are used for sentiment analysis, then accuracy can be maintained for small datasets, but processing time and labor requirements increase significantly for large volumes of data

Engineering Contradiction:

Improveprocessing speedVSAvoidtime required for manual screening

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent replaces manual screening and rule-based processing with a machine learning model that automatically performs sentiment analysis. The model processes unstructured text data autonomously, eliminating the need for human reviewers to manually screen documents while maintaining high processing speed and accuracy for large datasets

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The machine learning model performs self-learning and automatic classification of sentiment without requiring human intervention. The system processes unstructured text data independently, making decisions about aspect terms and sentiment classification autonomously based on trained patterns and features

Inventive Principle:

Principle #25Self-service

2Measurement precision

If manual customization and maintenance of rules is performed, then rule accuracy can be controlled, but the complexity and time required for processing increases

Engineering Contradiction:

Improverule accuracyVSAvoidcomplexity of rule set

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces manual rule customization and maintenance with an automated machine learning model. The model learns patterns from training data and automatically generates accurate sentiment classifications without requiring manual rule creation or maintenance, thereby reducing complexity while maintaining precision

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system transforms the approach from static manual rules to dynamic learned parameters. The machine learning model adapts its internal parameters based on training data, enabling it to accurately classify sentiment without requiring manual rule updates or maintenance, thus simplifying the overall system complexity

Inventive Principle:

Principle #35Parameter changes

3Productivity

If unstructured text data is processed using traditional methods, then data flexibility is maintained, but analysis efficiency decreases

Engineering Contradiction:

Improveanalysis efficiencyVSAvoidease of processing unstructured data

Core Design Contradiction:

ProductivityVSEase of operation

Solution Approach 1:

The patent replaces traditional manual processing methods with an automated machine learning model specifically designed to handle unstructured text data. The model efficiently processes and analyzes unstructured data without requiring manual intervention, thereby improving analysis efficiency while maintaining the flexibility to handle diverse text formats

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The machine learning model is designed to universally handle various types of unstructured text data across different domains and formats. The model can process diverse text structures and generate consistent sentiment analysis results, making the system both efficient and adaptable to different data types without requiring separate processing procedures

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS11675823B2Sentiment analysis for aspect terms extracted from documents having unstructured text data

Publication Date: 2023.06.13 DELL PROD LP

AI summary

An apparatus comprises at least one processing device configured to receive a query to perform sentiment analysis for a document, to generate, utilizing a first machine learning model, a first set of encodings classifying words of the document as being aspect or non-aspect terms, to generate, utilizing a second machine learning model, a second set of encodings classifying sentiment of the words of the document, and to determine, for a given aspect term, attention weights for a given subset of the words of the document surrounding the given aspect term. The processing device is also configured to generate, utilizing a third machine learning model, a given sentiment classification of the given aspect term based on the attention weights and a given portion of the second set of encodings for the given subset of the words, and to provide a response to the query comprising the given sentiment classification.