Word Vector Correction Using Dimensional Expansion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current natural language processing methods face challenges in accurately distinguishing between similar and non-similar words in vector spaces, leading to unclear distinctions in word similarity calculations.
Innovation Solution
A word vector correction method that expands n-dimensional word vectors to (n+1)-dimensional vectors, updates values based on similarity determination using synonym information, and applies linear discriminant analysis to disperse these values back into n-dimensional vectors, thereby adjusting distances between similar and non-similar words.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If word vectors are represented in n-dimensional space using standard methods, then computational efficiency is maintained, but the ability to clearly distinguish between similar and non-similar words deteriorates
Solution Approach 1:
The patent transforms n-dimensional word vectors into (n+1)-dimensional vectors by adding a new dimension that encodes similarity information. This dimensional expansion allows the vector space to capture both the original semantic relationships and explicit similarity metrics, enabling clearer distinction between similar and non-similar words while maintaining computational tractability through structured dimension addition.
2Measurement precision
If standard word vector methods are used, then processing speed is maintained, but the accuracy of similarity determination deteriorates
Solution Approach 1:
The patent performs preliminary similarity assessments using the extended (n+1)-dimensional vectors before conducting detailed analysis. The additional dimension pre-encodes similarity relationships, allowing the system to quickly filter and prioritize word pairs that require further processing, thereby reducing overall processing time while improving accuracy.
3Measurement precision
If the vector dimension is increased to improve similarity distinction, then measurement precision improves, but computational complexity increases
Solution Approach 1:
The patent enhances only the specific dimension necessary for similarity distinction (the (n+1)th dimension) while keeping the original n dimensions intact for their established purposes. This localized enhancement approach improves measurement precision for similarity assessment without unnecessarily increasing computational complexity across the entire vector space.
Data Source
AI summary
The present disclosure provides a word vector correction method using artificial intelligence technology. A word vector correction method using a word vector with n dimensions includes generating a first (n+1)-dimensional word vector using an average of elements included in a first n-dimensional word vector; generating a second (n+1)-dimensional word vector using an average of elements included in a second n-dimensional word vector; and determining whether a first word corresponding to the first word vector and a second word corresponding to the second word vector are similar to each other on the basis of specified synonym information.