Natural Language Understanding Feature Extraction Segmentation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Natural language understanding (NLU) systems face challenges in accurately determining meaning from text due to a large number of possible feature-value combinations, leading to decreased accuracy and overlap in meaning as the number of complex information targets increases.

Innovation Solution

The method involves processing text inputs using multiple statistical models associated with each feature, combining values to form complex information targets, and applying techniques like named entity processing and weighted mapping to enhance accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If the number of complex information targets is increased to handle more feature-value combinations, then the system can recognize more text meanings, but the accuracy of determining the correct meaning decreases

Engineering Contradiction:

Improvenumber of complex information targetsVSAvoidaccuracy of meaning determination

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent segments the complex information target determination process into multiple independent feature extraction stages. Each feature (intent, service, context) is extracted separately using dedicated statistical models, and then combined to form the complete meaning. This segmentation allows the system to handle many feature-value combinations while maintaining accuracy by processing each feature independently rather than as a single large classification problem.

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If more statistical models are used to process each feature independently, then the accuracy of value determination improves, but the device complexity increases

Engineering Contradiction:

Improveaccuracy of feature value determinationVSAvoidnumber of statistical models

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system divides the NLU task into separate feature extraction components, each with its own statistical model. This segmentation allows each model to specialize in extracting a specific feature (intent, service, or context), improving accuracy for each feature while keeping individual models manageable in size and complexity.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent employs multiple statistical models that can be selectively applied based on the feature being extracted. The same statistical modeling approach is universally applied across different features (intent, service, context), allowing the system to handle diverse feature types with a consistent methodology while maintaining flexibility in model selection and combination.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS9454525B2Information extraction in a natural language understanding system

Publication Date: 2016.09.27 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

A method of extracting information from text within a natural language understanding system can include processing a text input through at least one statistical model for each of a plurality of features to be extracted from the text input. For each feature, at least one value can be determined, at least in part, using the statistical model associated with the feature. One value for each feature can be combined to create a complex information target. The complex information target can be output.