Exploring Smartphone Keyboard Interactions for Experience Sampling Method driven Probe Generation
summary
Custom Android keyboard logs ~3,500 typing sessions from a 3-week in-the-wild study (N = 22) and labels probes as opportune (83%) or inopportune (17%). Sessions are represented as ITD sequences, converted with DFT to extract frequency features and paired with time-domain features (session length, duration, error rate). A generalized Random Forest, F-score 93% model predicts opportune probing moments; SHAP explainability shows session length and peak amplitude are the strongest contributors.
AI-written summary · about
related talks
24mSecurity at your fingertips: Theory → Practice!
21m“Trust Us”: Mobile Phone Use Patterns Can Predict Individual Trust Propensity
12mOn Suggesting Phrases vs. Predicting Words for Mobile Text Composition
06mExamining the Social Context of Alcohol Drinking in Young Adults with Smartphone Sensing
21mThe Impact of Word, Multiple Word, and Sentence Input on Virtual Keyboard Decoding Performance
05mFine-Grained and Context-Aware Behavioral Biometrics for Pattern Lock on Smartphones
11mBootstrapping User-Defined Body Tapping Recognition with Offline-Learned Probabilistic ...
13mUniKey: Enabling Surface-Based Typing with Commodity Smartwatches via Cross-Modal Learning