Password Autofill Form Classification for Accurate Input Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional password auto fill applications fail to accurately identify webpage forms or inputs, leading to incorrect recognition of login pages and password fields, resulting in inefficient and error-prone form filling.
Innovation Solution
An autofill module that identifies inputs of a form, applies an autofill template, detects errors, and updates the template based on correction indications, while using machine-learned models to determine form types and fill inputs accurately.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional password autofill applications use simple form identification methods, then the device complexity is low, but the measurement precision of identifying webpage forms and inputs deteriorates
Solution Approach 1:
The patent introduces an intermediary classification system between the autofill application and the webpage forms. This includes form classifiers that categorize forms into types (login, registration, contact, etc.) and input classifiers that identify specific input fields within forms. These intermediary classification layers enable accurate identification without requiring complex direct analysis of every webpage structure.
Solution Approach 2:
The patent segments the form identification process into distinct modular components: form detection modules that identify whether a page contains a form, form classification modules that categorize form types, and input classification modules that identify specific input fields. This segmentation allows each component to specialize in specific identification tasks, improving overall accuracy while maintaining manageable system complexity.
2Productivity
If conventional password autofill applications use manual form filling, then the device complexity is low, but the productivity of form filling deteriorates
Solution Approach 1:
The patent implements preliminary action by pre-classifying forms and inputs during browser navigation, before the user actually needs to fill them out. The system detects forms, classifies them into types, and identifies input fields in advance, storing this classification information for rapid autofill execution. This preliminary classification work eliminates the need for complex real-time analysis during the actual filling process.
Solution Approach 2:
The patent enables self-service through automated form detection, classification, and filling. The system automatically detects forms on webpages, classifies them without user intervention, identifies relevant input fields, and populates them with appropriate data. This automation eliminates manual form filling while the modular classification architecture keeps system complexity manageable.
3Reliability
If conventional password autofill applications lack error detection, then the device complexity is low, but the reliability of form filling deteriorates
Solution Approach 1:
The patent implements feedback mechanisms where the classification system continuously monitors form-filling accuracy. When classification errors are detected (such as misidentifying input fields or form types), the system receives feedback and adjusts its classification parameters accordingly. This feedback loop improves reliability over time while adding only moderate complexity through the incorporation of error detection and correction capabilities.
Data Source
AI summary
Embodiments relate to improved classification techniques for classifying form (e.g., webpage) forms and inputs for password auto fill applications. A module may detect a client device accessing a form (e.g., webpage) and may identify inputs of the form (e.g., webpage). The module may group two or more of the identified inputs into a bundle and apply the bundle to one or more machine learned bundle classifier models configured to determine a form type of the bundle. The module may also apply the bundle to the selected input classifier model. The module may automatically fill one or more inputs of the form (e.g., webpage) according to the input types identified by the selected input classifier model.