Application Classifier Using Latent Semantic Indexing Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing application search systems face challenges in identifying applications with similar functionalities due to the limitations of keyword-based search methods, which fail to accurately represent the conceptual similarity between applications.
Innovation Solution
The implementation of an application classifier using latent semantic indexing (LSI) feature vectors and a gradient-boosted decision tree model to determine the similarity between applications, trained with positive and negative exemplar applications, allowing for the classification of applications into categories based on their conceptual similarity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If keyword-based search methods are used to search for applications, then the search system is simple to implement, but the accuracy of identifying applications with similar functionalities deteriorates
Solution Approach 1:
The patent transforms the search approach from keyword-based text matching to vector-based semantic representation. Applications and search queries are converted into numerical vectors where dimensions represent semantic concepts, enabling similarity measurement through mathematical operations. This parameter transformation allows the system to capture functional similarities that keywords miss, resolving the contradiction between implementation simplicity and identification accuracy.
Solution Approach 2:
The patent replaces the mechanical keyword-matching system with a computational vector similarity system. Instead of searching for exact word matches, the system uses vector operations (dot products, cosine similarity) to measure semantic proximity. This substitution maintains computational efficiency while dramatically improving the accuracy of identifying functionally similar applications.
2Measurement precision
If latent semantic indexing and machine learning models are used to classify applications, then the accuracy of determining functional similarity improves, but the device complexity increases
Solution Approach 1:
The patent implements a two-stage approach where applications are pre-processed into semantic vectors during indexing, and then simple vector similarity operations are performed during search. The complex tasks of semantic representation and relationship learning are performed in advance during the vectorization and training phases, allowing the actual classification to rely on computationally efficient vector operations. This preliminary action reduces the complexity burden on the real-time classification system.
Solution Approach 2:
The patent introduces vector representations as an intermediary between raw application text and classification decisions. Instead of directly comparing application descriptions or using complex rule-based systems, the invention transforms text into numerical vectors that serve as mediators. These vectors capture semantic meaning and enable efficient similarity computation, simplifying the overall classification architecture while maintaining high accuracy.
Data Source
AI summary
An application classifier classifies applications using latent semantic indexing (LSI) vectors of the applications. The application classifier uses a machine-learned model generated based on pairs of LSI vectors of positive and negative training sets of applications, where the positive training set includes applications within a desired category and the negative training set includes applications outside of the desired category. For a given application, the application classifier determines whether the application belongs to the desired category based on similarity of an LSI vector of the application and LSI vectors of positive and negative exemplar applications, as determined by the machine-learned model. If the LSI vector of the application is similar to an LSI vector of at least one positive exemplar application and not similar to an LSI vector of any of the negative exemplar applications, the application is determined to belong to the desired category.