Participatory Sensing in Commerce: Using Mobile Phones to Track Market Price Dispersion Nirupama Bulusu (Portland State University) Chun Tung Chou, Salil Kanhere, Yifei Dong, Shitiz Sehgal, David Sullivan and Lupco Blazeski (University of New South Wales, Australia) Urban. Sense 08 11/08/2008
Price Dispersion “The empirical evidence for price dispersion in both online and offline markets is sizeable, pervasive and persistent” (Baye et al, Handbook of Economics and Information Systems, 2006) Attributed to “shoe leather” costs Urban. Sense 08 11/08/2008
Today Numerous on-line price comparison sites Shopzilla, Amazon, Froogle Information extraction from web databases easy to automate Price comparison sites for off-line markets too Prices from grocery shops manually copied in Hong Kong Petrol prices collected by volunteers or web site staff in US, UK, Australia Manual collection is cumbersome, errorprone and not up-to-date Urban. Sense 08 11/08/2008
Participatory Sensing to Track Price Dispersion Harness power of the collective via participatory sensing Consumers collect and share pricing information Design criteria: As automated as possible to reduce reluctance in participation Use camera phones to replace human sensing, processing and communication tasks Two proof-of-concept systems to demonstrate feasibility Mobi. Shop: Automated product price collection Petrol. Watch: Automated fuel price collection Urban. Sense 08 11/08/2008
Mobi. Shop System Architecture Request Internet GPRS/HSPDA/Wi. Fi Response Central Server Ma tch t ex dt loa Up d an ze aly ing Pro duc t Sea rch Sto res Qu ery
Mobi. Shop vs. Petrol. Watch Nearly identical system architectures Petrol. Watch – camera position important Special computer vision algorithms for extracting fuel price information (on server/camera phone) Use of GPS and GIS to simplify image processing Mobi. Shop Petrol. Watch Urban. Sense 08 11/08/2008
Open Problems Data integrity Privacy Statistical data perturbation, fudging data resolution etc. won’t suffice since individual data items are of interest here Anonymity Bad data discourages users, reputation ranking methods could compromise privacy and anonymity Require information flow to server without revealing identity Integrity, privacy and anonymity concerns are potential barriers to participation Incentive mechanism requires larger scale studies for validation Urban. Sense 08 11/08/2008
Related Work Mobile phones in e-commerce Rural microfinance (CAM) Fair trade (Reuters Market Light) Sensor Data Clearinghouses Sensor. Map, Sensor. Base Participatory Sensing Systems Agricultural price dissemination to farmers Diet. Sense, Traffic. Sense, Bike. Net, Cartel etc. Security and Privacy for Participatory Sensing Anony. Sense, Pool. View, Participatory Privacy Regulation Urban. Sense 08 11/08/2008
Conclusion Participatory Sensing to Track Market Price Dispersion Two proof-of-concept systems: Petrol. Watch and Mobi. Shop Addressed challenge of collecting non-structured information Addressed usability, cost barrier to participation Opportunities/Challenges Data Integrity, User privacy and Anonymity Tackling Other Barriers to Participation Through Incentives Augmentation of Geographic Information Systems Urban. Sense 08 11/08/2008
Скачать презентацию Participatory Sensing in Commerce Using Mobile Phones to
9d281931dcf8d459514ce8acb0485552.ppt