Autofill Field Classification via Secure Content Subsets

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing user interfaces on mobile devices are cumbersome for entering data, leading to inefficiencies and errors, especially with small screens, and current autofill solutions are unreliable, often requiring user intervention and compromising security by sending data to remote providers without proper consent.

Innovation Solution

A system that allows an operating system to send a secure subset of content to a remote provider for autofill requests, receiving suggested user values and field classification, and determining similarity metrics without storing user-inputted data, thus providing intelligent autofill suggestions while maintaining user privacy and security.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If autofill data is sent to remote providers for processing, then data entry efficiency is improved, but user data security deteriorates

Engineering Contradiction:
Improvedata entry efficiencyVSAvoiduser data security
Core Design Contradiction:
ProductivityVSObject-affected harmful factors

Solution Approach 1:

The operating system acts as an intermediary between the user device and remote providers. It processes and anonymizes data locally before sending to remote providers, and filters incoming autofill suggestions. This mediator role allows efficient remote processing while protecting user data security by preventing direct exposure of sensitive information.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If complete user data is transmitted to remote providers, then autofill accuracy is improved, but loss of information (privacy) worsens

Engineering Contradiction:
Improveautofill accuracyVSAvoiduser privacy
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The system extracts only the necessary identifying features from user data for remote processing, while keeping sensitive personal information local. The operating system sends anonymized datasets to remote providers for training and matching, achieving accurate autofill without transmitting complete user privacy information.

Inventive Principle:
Principle #2Taking out (Extraction)

3Measurement precision

If manual data entry is required, then data accuracy is improved, but productivity deteriorates

Engineering Contradiction:
Improvedata accuracyVSAvoiddata entry speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The system implements self-service through intelligent autofill that automatically completes forms based on learned patterns from user behavior. The operating system monitors typing patterns, form interactions, and data entry habits to provide accurate autocomplete suggestions, reducing manual input while maintaining data accuracy through user-specific learning.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4016371B1Systems and methods for autofill field classification
Publication Date: 2023.08.23 GOOGLE LLC
  • EP4016371B1 patent drawingFigure 1
  • EP4016371B1 patent drawingFigure 2
  • EP4016371B1 patent drawingFigure 3A

AI summary

An example method includes determining a subset of content excluding user-specific information displayed by an application on a user interface (UI) of a user device. The method further includes transmitting a request to a remote provider an autofill request for the application. The method also includes receiving one or more suggested user values and a field classification request for a text input field displayed by the application on the UI. The method further includes detecting a user input value in the text input field and determining a similarity metric for each of the one or more suggested user values based on a comparison of the user input value to each of the one or more suggested user values. The method also includes transmitting a response to the field classification request to the remote provide comprising the similarity metric for at least one of the one or more suggested user values.