Web Event Recording via Server-Side Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Web browsers that only support event bubbling struggle to observe user interface events that are consumed by applications without propagating them, leading to incomplete event streams, which hampers event handling and analysis.
Innovation Solution
A system and method that detects events in web browsers supporting both event capturing and bubbling, records event streams, and infers unobserved events by comparing event contexts, using a server to analyze and supplement missing events in browsers that only support event bubbling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If event capturing is used to observe all events, then complete event observation is achieved, but browser compatibility is reduced since only browsers supporting event capturing can observe unpropagated events
Solution Approach 1:
The patent introduces a server as an intermediary between browsers and event analysis systems. Browsers send event streams to the server, which then infers unobserved events using machine learning models trained on event context patterns. This mediator approach allows browsers without event capturing support to achieve complete event observation through the server's inference capabilities.
Solution Approach 2:
The system creates a virtual copy of the event stream by inferring missing events based on patterns learned from training data. Instead of directly capturing all events in the browser, the system reconstructs the complete event stream on the server by copying and supplementing the observed events with inferred ones, achieving completeness without requiring browser support for event capturing.
2Measurement precision
If event streams are recorded and analyzed to infer unobserved events, then event observation completeness is improved, but system complexity increases due to training data collection and machine learning model implementation
Solution Approach 1:
The system performs preliminary actions by collecting training data and training machine learning models in advance, before actual event inference is needed. This offline preparation phase creates reusable models that can then quickly infer unobserved events during runtime without requiring complex real-time processing, thereby reducing operational system complexity.
Solution Approach 2:
The system uses self-service by automatically collecting training data from event streams and using machine learning models to autonomously infer missing events. The inference process operates automatically without manual intervention, and the system continuously improves its inference accuracy by learning from accumulated data, reducing the need for manual system configuration and management.
Data Source
AI summary
Unobserved user interface events may be detected based on upon previously recorded data streams of events. The prior data streams are obtained by recording events from users who have browsers that support both event capturing and bubbling. When users with browsers that support only event bubbling interact with the page, the data stream is augmented by inferring unobserved events based on similarity to other event stream records.