WearMail: On-the-Go Access to Information in Your Email with a Privacy-Preserving Human ...
summary
WearMail provides a conversational watch and phone interface that retrieves specific items from personal email by recruiting crowd workers who never see raw content but instead generate obfuscated filters and example-based extractors. A user study of 200 search logs finds 250 long‑tail query types; an obfuscation experiment (24 queries, 10 workers each) shows 1000 words → ~80–100% email‑locating accuracy. Extraction builds web‑sourced regexes; in a 20‑item extraction test 59% return the correct item first and 35% second.
AI-written summary · about
chapters (9)
- 0:00Introduction to WearMail
- 2:22Problem with current assistants
- 3:27Privacy-preserving workflow
- 5:56Obfuscated metadata study
- 7:40Crowd programming extraction
- 9:47System performance evaluation
- 11:40Limitations and errors
- 12:22Summary and contributions
- 13:06Q&A session
related talks
19mNostrMail
12mInvisibility Cloak: Personalized Smartwatch-Guided Camera Obfuscation
15mFor 2 Years I'm the Only One Reading My Emails! (Almost)
13mBuilding Privacy Products: Field Notes
20mEnhancing Mobile Content Privacy with Proxemics Aware Notifications and Protection
26m"Reality Mining for Compan
49mBeyond Cookies: The Unseen Privacy Risks of Web Analytics
11mDistrust of big tech and a desire for privacy: Understanding the motivations...
