crowdsourcing using location
play

crowdsourcing using location- based questionnaires Google - PowerPoint PPT Presentation

From passive to active crowdsourcing using location- based questionnaires Google StreetView to rebuild a cyclists experience Mobile Ghent 13 Kevin Baker Kevin.baker@ugent.be Kevin.baker@routeyou.com Motivation and scope Motivation and


  1. From passive to active crowdsourcing using location- based questionnaires Google StreetView to rebuild a cyclist’s experience Mobile Ghent ’13 Kevin Baker Kevin.baker@ugent.be Kevin.baker@routeyou.com

  2. Motivation and scope

  3. Motivation and scope

  4. Motivation and scope • We already look at what people do ...

  5. Motivation and scope

  6. Study design • Proof-of-concept: ASKme application Location Question Lickert scale answers Skip or dropout Google Street View

  7. Study design • Under the hood • X,Y • Bearing • Pitch • Question • Possible Answers • Answer Type • Date And Time • Token • User Id • Given Answers

  8. Results • Automaticaly generated Q-locations based on road database • Question based on road attribution: • “is there a cycling lane visible?”, “Overall suitability of the road for cycling?”, “Overal saftey of road?”, “Is this road paved or unpaved?”, ... • All Centers of road segment longer than 500 m • ± 250 000 Questions / Belgium and The Netherlands

  9. Results Set 1: unordered list of locations and Set 2: spatially gridded list of locations questions and questions • 6915 answers • 3993 answers • march 2013 – june 2013 • july 2013 – now • 64% unique locations • 61% unique locations

  10. Results • 10-Questions-set • Goal: analyze spread among answers on same location • 10 Street View images querying the suitability for a cyclist • Timeframe: 7 days • 2 question sets; self-selection of type of cyclist: recreational or race • 118 unique questionnaire sessions • 65 recreational cyclists/32 race cyclists/21 both • Total of 139 answered question sets • 8.4 % dropout • ± 5 - 10 seconds per image

  11. Results Very Suitable Q X Y RTYscore Suitable 1 3.301907 50.87629 0.775 Moderatly Suitable 2 3.297142 50.8778 3.872 3 5.792186 50.75135 1.663 Less Suitable 4 3.768525 50.87891 0.775 Not Suitable 5 4.250293 50.85228 2.846 6 5.93369 50.38827 2.984 7 3.739064 51.04478 0.913 8 5.790763 50.75148 1.663 9 3.839237 50.87856 2.057 10 4.262866 51.11538 1.663

  12. Discussion and challenges • Feasability to cover a large geographic region • What do you really want to know and which information is redundant • Unique source of Volunteered Geographic information when combined with other sources • Remodeling RTY Formula for ‘nicest cycling routing’ • Point vs Sequence of points • People don’t like 2D, they want 3D/4D • Limits of the Google API

  13. Q&A Kevin Baker • Contacts: Kevin.baker@ugent.be Kevin.baker@routeyou.com • Website: www.routeyou.com • ASKme: www.routeyou.com/en/page/view/636/ask-me-expressing-your- opinion-on-the-bikeability-of-roads.en

  14. 1

  15. 2

  16. 3

  17. 4

  18. 5

  19. 6

  20. 7

  21. 8

  22. 9

  23. 10

Recommend


More recommend