extracting places from traces of locations
DESCRIPTION
Extracting Places from Traces of Locations. Paper Authors Jong Hee Kang Benjamin Stewart William Welbourne Gaetano Borriello. PowerPoint Author Michael Cook. Michael Cook. 4 th Year Computer Science (Junior) Co-oping at Synovus Interests Databases Networking Web Development - PowerPoint PPT PresentationTRANSCRIPT
Extracting Places from Extracting Places from Traces of LocationsTraces of Locations
Paper AuthorsPaper AuthorsJong Hee KangJong Hee Kang
Benjamin StewartBenjamin StewartWilliam WelbourneWilliam WelbourneGaetano BorrielloGaetano Borriello
PowerPoint AuthorPowerPoint AuthorMichael CookMichael Cook
Michael CookMichael Cook
44thth Year Computer Science (Junior) Year Computer Science (Junior) Co-oping at SynovusCo-oping at Synovus InterestsInterests
DatabasesDatabases NetworkingNetworking Web DevelopmentWeb Development
Twin brotherTwin brother
The ProblemThe Problem
Location aware systems today are Location aware systems today are limitinglimiting
Place: An area of importance to a Place: An area of importance to a useruser
Usage Example:Usage Example: Cell phone goes to “silent” mode when Cell phone goes to “silent” mode when
entering a classroomentering a classroom
Ideal SituationIdeal Situation
Requires little user interactionRequires little user interaction All important places are locatedAll important places are located No false positivesNo false positives Works for indoor and outdoor placesWorks for indoor and outdoor places
Tracking User MovementTracking User Movement
Place Lab access Place Lab access pointspoints
Works indoorsWorks indoors
Popular Clustering Popular Clustering AlgorithmsAlgorithms
K-meansK-means Gaussian mixture modelGaussian mixture model
Large amounts of computationLarge amounts of computation
Time-Based ClusteringTime-Based Clustering
Streaming Streaming computationcomputation
Small clusters Small clusters ignoredignored
Time threshold and Time threshold and distance threshold distance threshold can be changedcan be changed
Time-Based Clustering Time-Based Clustering ResultResult
Changing Distance and Changing Distance and TimeTime
Changing Distance and Changing Distance and TimeTime
d=30m t=300secd=30m t=300sec d=50m t=300secd=50m t=300sec d=300m t=600secd=300m t=600sec
Frequently Visited PlacesFrequently Visited Places
Not much time is spent at the place, Not much time is spent at the place, but frequently visitedbut frequently visited
Different time threshold neededDifferent time threshold needed How to differentiate the place and in-How to differentiate the place and in-
transit motion?transit motion?
Future WorkFuture Work
Automatic labeling of placesAutomatic labeling of places Can use user’s calendarCan use user’s calendar
Learn proper distance and time Learn proper distance and time thresholds automaticallythresholds automatically
CritiqueCritique
Easy to read and Easy to read and understandunderstand
Cool idea with Cool idea with practical practical applicationsapplications
WiFi hotspots not WiFi hotspots not always availablealways available
Trying to do too Trying to do too much at oncemuch at once Long duration Long duration
placesplaces Short duration, Short duration,
frequent placesfrequent places
Questions?Questions?