Semantic Vector Generation for Multisense Word Disambiguation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional techniques for computing word vectors, such as Word2Vec, fail to distinguish multiple semantics of a multisense word, resulting in low relevance semantic vectors due to their reliance on co-occurrence relations alone.

Innovation Solution

A computer-readable storage medium with a program that generates semantic vectors by combining vectors of synonyms and feature words from a storage unit, allowing for the creation of vectors specific to each semantics of a multisense word.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If a single vector is assigned to a multisense word based on co-occurrence relations, then the computation process is simple, but the semantic relevance to each specific meaning is low

Engineering Contradiction:

Improvecomputation simplicityVSAvoidsemantic relevance

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent segments the single vector representation into multiple vectors, each corresponding to a specific semantics of the multisense word. By dividing the vector space according to different meanings, the system achieves both computational feasibility and semantic precision for each segmented meaning.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by assigning different vector characteristics to different semantics of the same word. Each semantics receives a vector with specific properties tailored to its meaning, rather than using a uniform vector for all meanings, thereby improving semantic relevance locally for each meaning.

Inventive Principle:

Principle #3Local quality

2Measurement precision

If multiple vectors are generated for each semantics of a multisense word, then the semantic representation accuracy is improved, but the computational complexity increases

Engineering Contradiction:

Improvesemantic representation accuracyVSAvoidcomputation complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary action by pre-identifying the semantics of multisense words and pre-organizing the vector generation process for each semantics. This preparation work is done before the actual vector computation, which streamlines the subsequent processing and reduces overall computational complexity despite generating multiple vectors.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent creates a universal framework that can handle multiple semantics of a word using the same computational approach. The vector generation mechanism is designed to be multi-functional, capable of producing vectors for different semantics through a unified process, thereby managing complexity while maintaining accuracy.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS11514248B2Non-transitory computer readable recording medium, semantic vector generation method, and semantic vector generation device

Publication Date: 2022.11.29 FUJITSU LTD

AI summary

A semantic vector generation device (100) obtains vectors of a plurality of words included in text data. The semantic vector generation device (100) extracts a word included in any group. The semantic vector generation device (100) generates a vector in accordance with the any group on the basis of a vector of the word extracted among the obtained vectors of the words. The semantic vector generation device (100) identifies a vector of a word included in an explanation of any semantics of the word extracted among the obtained vectors of the words. The semantic vector generation device (100) generates a vector in accordance with the any semantics on the basis of the vector identified and the vector generated.