ML Hypertext Element Classification via Token Embeddings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hypertext analysis techniques, such as autofillers, face challenges in accurately classifying and filling out forms and fields in HTML pages due to limitations in processing raw HTML files and performing specific tasks like form and field analysis.
Innovation Solution
The proposed system employs a machine learning (ML) model, pre-trained on hypertext data and fine-tuned for specific classification tasks, to generate contextualized vector representations of hypertext elements. These representations are used to classify elements into specific categories, such as forms and fields, enabling more accurate autofilling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional scanner tools are used to analyze hypertext for form filling, then the system can process HTML pages, but the classification accuracy of forms and fields is insufficient
Solution Approach 1:
The patent transforms the input hypertext into multiple parameter representations including tokens, positional embeddings, and path embeddings. These different parameter representations are then processed by the machine learning model to achieve accurate classification of form elements, directly resolving the contradiction between processing capability and classification accuracy
Solution Approach 2:
The system creates a composite representation by combining multiple types of embeddings (tokens, positional, path) to form a comprehensive feature vector for each form element. This composite approach enables the model to capture both content and structural information, improving classification accuracy and subsequent form filling reliability
2Measurement precision
If a machine learning model is used to generate contextualized vector representations, then classification accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the complex task of form element classification into distinct processing stages: token generation, positional encoding, path encoding, and final classification. This segmentation allows each component to be optimized independently while maintaining overall system accuracy, managing computational complexity through modular design
Solution Approach 2:
The machine learning model serves as an intermediary that transforms raw hypertext inputs into contextualized vector representations. This intermediary layer enables accurate classification without requiring direct complex processing of the entire HTML structure, simplifying the overall system architecture while maintaining high accuracy
3Manufacturing precision
If multiple types of tokens are generated and processed, then classification precision is enhanced, but processing time increases
Solution Approach 1:
The system performs preliminary processing by generating all necessary token representations (tokens, positional embeddings, path embeddings) before the classification stage. This preliminary action organizes the data in advance, allowing the machine learning model to process pre-structured information efficiently, reducing actual classification time while maintaining high precision
Data Source
AI summary
The present application describes systems and methods for analyzing hypertext. An element detector receives input information (e.g., hypertext) that is included in a web page. The element detector identifies one or more elements in the input information that are relevant to a classification task. The classification task may include classifying the one or more elements. An element extractor generates text tokens, positional tokens, and/or path tokens based on the identified one or more elements. A machine learning (ML) model outputs one or more vector representations based on the text tokens, positional tokens, and/or path tokens. An output classifier, which may be an additional layer of the ML model, classifies the one or more elements into classifications based on the one or more vector representations. A tool may perform one or more tasks based on the classifications.