ERICA: Interaction Mining Mobile Apps
summary
ERICA is a web-based system that mines live Android apps without modifying them, recording screenshots at >15 fps, fast view hierarchies (≈0.1 s dumps) and user events to produce unified interaction traces. It collects traces from over 1,000 apps (≈18,000 unique screens, 52,000 interactions, ~6M elements), trains classifiers to detect 23 flow types and indexes ~3,000 examples, and exposes an online flow search. The pipeline runs with sub‑second latency and supports motion, element-level indexing and reverse‑engineering use cases.
AI-written summary · about
chapters (11)
- 0:00Introduction to mobile app design
- 1:10Data-driven design tools
- 2:56Introducing ERICA system
- 5:08How ERICA captures data
- 7:16Combining data streams
- 8:22Data collection results
- 9:44Limitations of existing repositories
- 10:47Identifying user flows
- 11:29Flow search engine demo
- 12:31Summary and future work
- 14:58Q&A session
related talks
29mWhy ANRs are Hard to Solve and What You Can Do to Make it Easier
40mAndroid performance monitoring with automation
20mZIPT: Zero-Integration Performance Testing of Mobile App Designs
43mAndroid Dev Tools and Workflows
19mSolve your Mobile Bugs with Session Replay
51mAndrosia: A step ahead in securing in-memory Android application data
20mDemystifying the Android Telehealth Ecosystem
34mAgile UX Research Practice in Android