Trace2TAP: Synthesizing Trigger-Action Programs From Traces of Behavior
summary
Method synthesizes all trigger-action programs that explain past manual actuations using symbolic reasoning and an SMT solver, handles out-of-order events and proposes repairs when users revert automations, implements on Samsung SmartThings and adds a discrete k-means clustering and ranking plus visualization. Field deployments collect >1M events from 53 devices across formative and field studies, target 18 actions, and users pick 32 synthesized rules from 31 clusters (median rank 2), with most actions finding usable automations.
AI-written summary · about
chapters (9)
- 0:00Intro
- 0:14Developing Automations in Smart Homes
- 1:13Key Challenge: Trace + Intended Automation
- 1:51Key Challenge No Single Criterion Defines the "Best" Rule
- 2:05Limitations of Previous Work
- 2:27Trace2TAP: Our Solution
- 2:55Step 1: Rule Synthesis
- 4:00Step 2: Rule Presentation (Clustering)
- 4:41Evaluation
related talks
21mTransactuations: Where Transactions Meet the Physical World
51mCreate IoT Automation Using SmartThings New Rules API
17mSensitive Information Tracking in Commodity IoT
23mAcquisitional Rule-based Engine for Discovering Internet-of-Things Devices
29mSmartAuth: User-Centered Authorization for the Internet of Things
20mEmpowering End Users to Customize their Smart Environments: Model, Composition Paradigms, and ...
44mScaling Your Business with SmartThings
39mUnleash the Power of Cloud Connected Devices with SmartThings