Text Object Classification Verification Using Embeddings And Binary Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to accurately detect and correct misclassifications in large datasets of text-based objects, making it difficult to maintain accurate categorization schemes.
Innovation Solution
A class verification platform utilizing vector embeddings and a binary search process to determine similarity scores between subject and class text strings, recursively splitting and reclassifying misclassified text strings, and employing clustering to correct inaccuracies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If vector embeddings and binary search process are used to detect misclassifications, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent segments the classification verification process into multiple stages: (1) generating vector embeddings for subject and class text strings, (2) computing similarity scores, (3) recursively splitting subject text strings into subsets, and (4) reclassifying based on similarity thresholds. This segmentation allows complex misclassification detection to be broken down into manageable steps, improving precision while controlling complexity through structured processing.
Solution Approach 2:
The patent transforms text strings into vector embeddings, moving from traditional string-based classification to a dimensional vector space. This dimensional transformation enables more nuanced similarity measurements and improves classification accuracy by capturing semantic relationships that text-based methods alone cannot detect.
2Measurement precision
If recursive binary search process is applied to split and reclassify text strings, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent performs preliminary actions by pre-computing vector embeddings for text strings and pre-establishing similarity thresholds before actual misclassification detection is needed. This allows the recursive binary search process to operate more efficiently during execution, reducing the time required for each classification verification step while maintaining high precision.
Solution Approach 2:
The patent implements a dynamic recursive binary search process that adapts its behavior based on similarity scores and threshold comparisons. The process dynamically splits and merges text string subsets based on real-time similarity measurements, optimizing processing time by focusing computational resources only on portions of the data that require reclassification.
3Measurement precision
If text strings are split into multiple subsets for reclassification, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent segments text strings into multiple subsets based on similarity score thresholds and recursively processes each subset separately. This segmentation strategy improves precision by allowing targeted reclassification of only those text strings that fail the initial similarity test, while managing processing complexity through systematic division of the data into manageable groups.
Data Source
AI summary
In some aspects, the techniques described herein relate to a method including: receiving, as input to a binary search process, a subject vector embedding and a class vector embedding, wherein the subject vector embedding is generated from a plurality of subject text strings and wherein the class vector embedding is generated from a class text string; generating a similarity score; determining that the similarity score is below a threshold value; splitting the plurality of subject text strings into a first new plurality of subject text strings and a second new plurality of subject text strings; receiving a new subject vector embedding, wherein the new subject vector embedding is generated from the first new plurality of subject text strings; and calling the binary search process recursively using the new subject vector embedding and the class vector embedding as input to the binary search process.