Crowd-Sourced UI Test Generation for Enterprise Mobile Apps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for automatically generating user interface test cases for enterprise mobile applications are inaccurate in representing real-world user actions and struggle to identify the most useful test cases due to limitations in resources and the synthetic nature of model-generated test cases.
Innovation Solution
The method involves downloading a mobile application to multiple devices, capturing session logs of user interactions, and analyzing these logs to discover patterns that occur frequently across sessions, thereby generating new test cases that accurately reflect real-world user actions and are relevant to specific contexts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If machine learning models are used to automatically generate test cases from crowd-sourced data, then test case generation is automated, but the accuracy of representing real-world user actions deteriorates
Solution Approach 1:
The patent creates actual copies of real user sessions by recording and replaying authentic user interactions. Instead of using synthetic model-generated test cases, the system captures real user actions, device states, and contextual data, then uses these actual copies as test cases. This ensures 100% fidelity to real-world behavior while maintaining full automation through the recording and replay mechanism.
Solution Approach 2:
The system enables test cases to generate themselves automatically through the natural usage of the application by real users. As users interact with the app, their sessions are automatically recorded and converted into test cases without requiring manual intervention or external model generation. The real user behavior itself serves as the source material for creating the test suite.
2Reliability
If thousands of test cases are created to cover all user actions, then test coverage is improved, but resource consumption and testing time increase
Solution Approach 1:
The patent applies partial action by selecting and executing only the most relevant test cases from the collected session data. Rather than running all possible test cases, the system identifies high-value test cases based on frequency, criticality, and coverage metrics, executing only those necessary to achieve sufficient testing coverage. This reduces testing time while maintaining adequate reliability.
Solution Approach 2:
The test suite is segmented into priority levels and execution batches. Test cases are divided into critical, important, and optional categories, allowing the system to execute high-priority cases first and stop when coverage thresholds are met. This segmentation enables efficient resource allocation and reduces overall testing time while maintaining comprehensive coverage of critical functionality.
3Measurement precision
If manual test case writing is performed to ensure accuracy, then test case quality improves, but time consumption and difficulty increase
Solution Approach 1:
The system creates exact copies of real user sessions, capturing authentic user actions, device states, and contextual information. These copies serve as high-quality test cases that inherently reflect real-world usage patterns without requiring manual writing or interpretation. The copying process automatically preserves all nuances of actual user behavior.
Solution Approach 2:
Test cases are self-generated from actual user interactions with the application. As users naturally use the app, their sessions are automatically recorded and converted into executable test cases. This self-service mechanism eliminates the need for manual test case writing while ensuring high quality through direct observation of real user behavior.
Data Source
AI summary
A computer-implemented method includes downloading respective instances of an enterprise mobile application to a plurality of mobile devices. The instances of the enterprise mobile applications, while executing on respective mobile devices, capture, for each session, a session log that includes indications of ordered user actions occurring during the session, and optionally time intervals between user actions and/or user attributes. Captured session logs stored at and are mined by one or more servers to discover a particular pattern or sequence of user actions that occurred across multiple, different user sessions. If the number and/or rate of occurrences of the particular pattern is greater than a threshold, a new test case corresponding to the pattern is automatically generated and added to a suite of test cases for the UI functionality of the enterprise mobile application. The updated test suite may be automatically executed on a test version of the enterprise mobile application.