際際滷

際際滷Share a Scribd company logo

Photographer Paths: Sequence Alignment of Geotagged Photos for Exploration-based Route Planning

Abdallah El Ali
Abdallah El Ali

際際滷s for the talk I gave at CSCW 2013, held in San Antonio, TX, USA. The full paper reference is: El Ali, A., van Sas, S. & Nack, F. (2013). Photographer Paths: Sequence Alignment of Geotagged Photos for Exploration-based Route Planning. In proceedings of the 16th ACM Conference on Computer Supported Cooperative Work and Social Computing (CSCW '13), 2013, San Antonio, Texas. Paper link: http://staff.science.uva.nl/~elali/pdfs/p985-el-ali.pdf

1 of 44
Downloaded 13 times
Photographer Paths: Sequence Alignment of Geotagged Photos for Exploration-based Route Planning! Feb. 油26, 油2013 Abdallah 油Abdo 油El 油Ali 油 Sicco 油van 油Sas 油 Frank 油Nack 油 h6p://sta鍖.science.uva.nl/~elali/ 油
Outline! I. Introduc3on 油 II. Photographer 油Paths 油 III. User 油Evalua3on 油 IV. Results 油 V. Discussion 油& 油Future 油Work 油 2
Introduction 3
4
5
Pffttt 6
油 油We 油dont 油always 油want 油to 油 supply 油user 油preferences 油 7
油Social 油and 油local 油interpreta3on 油of 油city 油places 油and 油routes 油 8
O鍖-足the-足beaten 油track, 油social 油trails, 油 9 油 油Lonely 油Planet! 油 油
油Assump3on: 油 油Loca3ons 油of 油photographs 油are 油poten3ally 油interes3ng 油 10
油But 油sequen3al 油property 油needs 油to 油be 油captured! 油 11
油Sequence 油Alignment 油methods 油 12
油Analysis 油of 油mobility 油behavior 油of 油city 油 photographers: 油 р where 油photographers 油have 油been 油 р in 油what 油order 油they 油have 油been 油 there 油 р how 油closely 油their 油movements 油 parallel 油those 油of 油other 油 photographers 油 13 By Keiichi Matsuda via supercolossal
Research Questions! 油How 油can 油walkable 油route 油plans 油be 油automaCcally 油generated 油for 油residents 油(and 油tourists) 油 that 油would 油like 油to 油explore 油a 油city? 油 油And 油are 油these 油route 油plans 油desirable? 油 油Three 油factors: 油 油1) 油Which 油data 油sources? 油 油2) 油Which 油methods 油to 油generate 油routes? 油 油 油 油 油 油 油 油 油 油 油3) 油User 油percep3ons 油compared 油to 油fastest 油 and 油popular 油routes? 油 油 14
Photographer Paths! 15
Approach! 油1) 油Crawl 油Flickr 油 油geotags, 油3mestamps 油 油2) 油Map 油each 油geotag/loca3on 油in 油a 油sequence 油to 油a 油cell 油in 油a 油par33oned 油grid 油map 油 油3) 油Mul3ple 油Sequence 油Alignment 油on 油photographer 油routes 油to 油鍖nd 油aligned 油loca3on 油sequences 油 油 油These 油alignments 油are 油Photographer 油Route 油Segments 油(PRSs) 油 16
Dataset! р Flickr 油geotagged 油photos 油within 油Amsterdam, 油The 油Netherlands 油 油 р Area: 油17.3 油km 油N-足S 油and 油24.7 油km 油E-足W 油(center) 油 р 5-足year 油period 油(Jan. 油2006 油-足 油Dec. 油2010) 油 р Aeributes: 油 油 р owner 油ID 油 р photo 油ID 油 р date 油and 油3me-足stamp 油 р la3tude 油and 油longitude 油(street 油level 油accuracy) 油 油Database: 油426,372 油photos 油 17
Preprocessing! р Sequence 油inference 油with 油following 油constraints: 油 р photo 油taken 油within 油4 油hours 油from 油previous 油photo 油and 油in 油same 油order 油 р minimum 油2 油or 油more 油di鍖erent 油loca3ons 油(or 油nodes) 油 р early 油experiments 油determined 油125 油x 油125m 油cells 油in 油center 油of 油Amsterdam 油grid 油suitable 油 油1691 油routes 油(average 油length 油of 油9.92 油loca3ons) 油 油1130 油unique 油photographers 油 18
Sequence Alignment! р To 油鍖nd 油photographer 油paths 油from 油Photographer 油Route 油Segments 油(PRSs), 油constraints 油set: 油 р PRS 油has 油minimum 油4 油photographers 油with 油minimum 油2 油aligned 油nodes/loca3ons 油 !#**% Photographer Route Segments (PRSs)! !"#$% !"**% !+**% %&'()*+$,*-.*)/*0$ !***% $**% ##&% +%,-./.012,-314% 油 油 油 油231 油PRSs 油 油 油 油 #**% '!&% )%,-./.012,-314% 油(average 油length 油of 油2.61 油nodes) 油 "%,-./.012,-314% "**% !"#$ +**% !'(% !*!% +'% !*% !% )'% *% +% )% "% '% 1)'-.*$&2/342)0$ 19
PRSs in Amsterdam! 20
PRS Aggregation! р Modi鍖ed 油Dijkstras 油shortest 油path 油algorithm 油 & P1N1 P2N1 & # ! P1N2 ! " # !" #$% Start P1N3 P2N2 End 21
PRS Aggregation to Crude Routes! PRSs CM Photographer Route WW Photographer Route 22
User Evaluation 23
Laboratory Study Design! р ~45 油min. 油Quan3ta3ve/Qualita3ve 油lab-足based 油study 油 р 15 油par3cipants 油(10 油m, 油5 油f) 油aged 油between 油21-足35 油(M 油= 油29.2; 油SD 油= 油3.3) 油 р Interac3ve 油web-足based 油prototype 油route 油planner 油 р Expert 油route 油evalua3on 油by 油city 油residents 油(lived 油in 油Amsterdam 油> 油1 油year) 油 р Plain 油routes 油to 油avoid 油informa3on 油type 油bias 油 24
Laboratory Study Design! р Two 油scenarios: 油 р Route 油1: 油Central 油Sta3on 油to 油Museumplein 油 油anernoon 油scenario 油favoring 油explora3on 油 р Route 油2: 油Waterlooplein 油to 油Westerkerk 油 油evening 油scenario 油favoring 油e鍖ciency 油 р Baseline 油comparisons: 油 油 р Photo 油Density 油(PD) 油route: 油highest 油density 油of 油photos 油(over 油5 油year 油period) 油in 油grid 油cells 油 along 油route 油 油 р Google 油Maps 油(GM) 油route: 油shortest 油route 油between 油two 油loca3ons 油 р Counterbalanced 油within-足subject 油design 油 油 р Route 油Varia3on 油(IV): 油Photographer 油Paths 油vs. 油Photo 油Density 油vs. 油Google 油Maps 油 25
Central Station to Museumplein (CM)! Photographer Paths Photo Density Google Maps route (5.36 km) route (3.83 km) route (3.35 km) 26
Waterlooplein to Westerkerk (CM)! Photographer Paths route (2.28 km) Photo Density route (2.60 km) Google Maps route (1.59 km) 27
Laboratory Study Design! 油Data 油collected: 油 1. AerakDi鍖2 油(Hassenzahl, 油2003) 油UX 油ques3onnaire 油responses 油[7-足point 油seman3c 油 di鍖eren3al 油scale]: 油Usability, 油Hedonic 油Quali3es 油(Iden3ty, 油S3mula3on), 油Aerack3veness 油 2. Two-足part 油semi-足structured 油interviews 油 油Part 油1: 油Route 油preferences, 油feedback 油on 油Photographer 油Paths 油 油Part 油2: 油Inves3ga3on 油of 油visualized 油informa3on 油types 油(visualized 油info 油type 油handouts): 油 油 油a) 油Google 油maps 油 油 油b) 油Color 油coded 油PRSs 油(PP 油route) 油 油 油c) 油Density 油geopoints 油(PD 油route) 油 油d) 油Thumbnail 油photo 油geopoints 油 油 油e) 油Foursquare 油POIs 油 28
a) d) b) e) c) 29
Web Survey Study! р Short 油web-足based 油survey 油for 油CM 油and 油WW 油routes 油and 油varia3ons 油 р Basic 油demographics 油collected: 油age, 油gender, 油years 油in 油Amsterdam 油 р Sta3c 油route 油images, 油no 油counterbalancing 油 р 82 油par3cipants 油(55 油m, 油27 油f) 油aged 油between 油17-足62 油(M= 油27.6; 油SD= 油6.1) 油 油 р Most 油lived 油in 油Amsterdam 油for 油more 油than 油3 油years 油(44/82) 油 油 р Some 油between 油1-足3 油years 油(15/82) 油 р Less 油than 油a 油year 油(11/82) 油 油 р Only 油visited 油before 油(12/82) 油 30
Results 31
AttrakDiff2 ! Central Station to Museumplein (CM) Route "# ** ** ** ** * ** ** * $# %# !"#$% &# !%# !$# !"# '()*+),-#./)0123#4'.5# 6789:1-#./)0123#!#;87:,23# 6789:1-#./)0123#!# =>()-,?7:7@@#4=AA5# 46.!;5# <,+/0),9:#46.!<5# &'("$)'*$% ** !"#"$%$& * !"#"$%$' 'B929*()CB7(#')2B@# 'B929#D7:@123# E99*07#F)C@# 32
AttrakDiff2 ! Waterlooplein to Westerkerk (WW) Route ** ** "# ** ** $# %# !"#$% &# !%# !$# !"# '()*+),-#./)0123#4'.5# 6789:1-#./)0123#!#;87:,23# 6789:1-#./)0123#!# =>()-,?7:7@@#4=AA5# 46.!;5# <,+/0),9:#46.!<5# &'("$)'*$% !"#"$%$& ** 'B929*()CB7(#')2B@# 'B929#D7:@123# E99*07#F)C@# 33
Route Preference! Lab 油Study 油 р CM 油route: 油most 油chose 油to 油follow 油the 油PP 油route 油(9/15), 油PD 油route 油(4/15), 油GM 油route 油(2/15) 油 р One 油of 油the 油routes 油[PP] 油was 油long 油and 油took 油many 油detours, 油and 油I 油thought 油that 油was 油a 油 very 油aFracHve 油route! 油 р WW 油route: 油most 油chose 油to 油follow 油GM 油route 油(10/15), 油PD 油route 油(4/15), 油no 油route 油(1/15) 油 油 р You 油are 油going 油for 油co鍖ee 油so 油you 油just 油want 油to 油get 油there, 油unlike 油in 油the 油鍖rst 油[CM] 油 scenario 油where 油it 油is 油a 油nice 油day 油and 油you 油have 油Hme. 油 Web 油Survey 油 ≒ 油 油 油 油 油CM 油route: 油GM 油(40/82), 油PD 油(23/82), 油PP 油route 油(10/82), 油neither 油(9/82) 油 ≒ 油No 油experimenter 油steering; 油many 油Amsterdam 油residents 油know 油the 油city 油already 油quite 油 well! 油 ≒ 油I 油would 油not 油easily 油walk 油these 油routes... 油who 油in 油Amsterdam 油walks? 油;) 油 ≒ 油 油 油 油 油WW 油route: 油GM 油(67/82), 油PD 油(6/82), 油PP 油route 油(3/82), 油neither 油(6/82) 油 34
Digital Information Aids! Lab 油Study 油 油 油 油 油 油 油 油Interview: 油Part 油1 油 р How 油many 油persons 油(focus 油on 油city 油photographers) 油took 油a 油given 油route 油segment 油over 油a 油certain 油3me 油 period 油(e.g., 油1 油year)? 油 р Useful 油(8/15) 油for 油exploring 油a 油city 油one 油already 油knows 油 р Not 油sure 油(4/15) 油 р Depends 油on 油which 油photographers 油(2/15) 油 р Not 油for 油me 油(1/15) 油 油Interview: 油Part 油2 油 р Found 油PP 油info 油type 油aerac3ve 油(10/15), 油but 油combine 油with 油Photo 油thumbnails 油(3/10) 油and 油POIs 油(3/10) 油 35
Digital Information Aids! Web 油Survey 油 р POIs 油along 油a 油route 油(51x) 油 р Route 油distance 油(51x) 油 р Comments 油along 油a 油route 油(ranked 油by 油highest 油ra3ngs 油or 油recency) 油(24x) 油 р Expert 油travel 油guides 油(22x) 油 р Photos 油of 油route 油segments 油(17x) 油 р No 油digital 油aids 油(13x) 油 р Number 油of 油photographers 油that 油took 油a 油given 油path 油over 油a 油Hme 油period 油(9x) 油 р Number 油of 油photos 油along 油a 油route 油over 油a 油3me 油period 油(9x) 油 36
Discussion 37
Discussion! р Discrepancy 油between 油lab-足study 油and 油web 油survey 油 р Quick 油web 油survey 油insu鍖cient? 油 油 р Visualiza3on/explana3on 油of 油digital 油aids 油important? 油 р Proof-足of-足concept 油approach 油requires 油real-足world 油outdoor 油evalua3on 油 р Di鍖erent 油street 油grid 油network 油 р Scalability 油to 油larger 油ci3es 油 р More 油context-足awareness 油 38
Take Home Message! Going 油towards 油data-足driven 油explora3on-足based 油route 油planners 油 р Some3mes 油its 油the 油journey, 油not 油the 油des3na3on 油 р A 油quan3ta3ve 油approach 油may 油oversimplify 油human 油needs 油for 油explora3on 油 р But 油some3mes 油we 油want 油an 油automa3c 油solu3on, 油so 油as 油not 油to 油be 油bothered 油 with 油supplying 油user 油preferences 油and 油encounter 油serendipity 油 39
Questions 40 h6p://sta鍖.science.uva.nl/~elali/ 油
References! 油1. 油Cheng, 油A.-足J., 油Chen, 油Y.-足Y., 油Huang, 油Y.-足T., 油Hsu, 油W. 油H., 油and 油Liao, 油H.-足Y. 油M. 油Personalized 油travel 油 recommenda3on 油by 油mining 油people 油aeributes 油from 油community-足contributed 油photos. 油In 油Proc. 油 MM 油11, 油ACM 油(2011), 油8392. 油 油2. 油M. 油Clements, 油P. 油Serdyukov, 油A. 油P. 油de 油Vries, 油and 油M. 油J. 油Reinders. 油Using 油鍖ickr 油geotags 油to 油predict 油 user 油travel 油behaviour. 油In 油Proc. 油SIGIR 油10, 油pages 油851852. 油ACM 油Press, 油2010. 油 3. 油M. 油De 油Choudhury, 油M. 油Feldman, 油S. 油Amer-足Yahia, 油N. 油Golbandi, 油R. 油Lempel, 油and 油C. 油Yu. 油 Automa3c 油construc3on 油of 油travel 油i3neraries 油using 油social 油breadcrumbs. 油In 油Proc. 油HT 油10, 油pages 油 3544. 油ACM 油Press, 油2010. 油 4. 油F. 油Girardin, 油F. 油Calabrese, 油F. 油D. 油Fiore, 油C. 油Ra|, 油and 油J. 油Blat. 油Digital 油footprin3ng: 油Uncovering 油 tourists 油with 油user-足generated 油content. 油IEEE 油Pervasive 油Compu3ng, 油7:3643, 油October 油2008. 油 5. 油N. 油Shoval 油and 油M. 油Isaacson. 油Sequence 油alignment 油as 油a 油method 油for 油human 油ac3vity 油analysis 油in 油 space 油and 油3me. 油Annals 油of 油the 油Associa3on 油of 油American 油Geographers, 油97(2):282297, 油2007. 油 6. 油A. 油Vaccari, 油F. 油Calabrese, 油B. 油Liu, 油and 油C. 油Ra|. 油Towards 油the 油socioscope: 油an 油informa3on 油system 油 for 油the 油study 油of 油social 油dynamics 油through 油digital 油traces. 油In 油Proc. 油GIS 油09, 油pages 油5261. 油ACM 油 Press, 油2009. 油 7. 油Hassenzahl, 油M., 油Burmester, 油M., 油and 油Koller, 油F. 油AerakDi鍖: 油Ein 油Fragebogen 油zur 油Messung 油 wahrgenommener 油hedonischer 油und 油pragma3scher 油Qualit即at. 油Mensch 油& 油Computer 油2003. 油 Interak3on 油in 油Bewegung 油(2003), 油187196. 油 8. 油Lu, 油X., 油Wang, 油C., 油Yang, 油J.-足M., 油Pang, 油Y., 油and 油Zhang, 油L. 油Photo2trip: 油genera3ng 油travel 油routes 油 from 油geo-足tagged 油photos 油for 油trip 油planning. 油In 油MM 油10, 油ACM 油(2010), 油143152. 油 9. 油Wilson, 油C. 油Ac3vity 油paeerns 油in 油space 油and 油3me: 油calcula3ng 油representa3ve 油hagerstrand 油 trajectories. 油TransportaHon 油35 油(2008), 油485499. 油 41
Related Work! р Sequence 油Alignment 油(SA) 油methods: 油 р Borrowed 油from 油bioinforma3cs 油and 油later 油3me 油geography 油 р Time 油geography 油systema3cally 油analyzes 油and 油explores 油the 油sequen3al 油dimension 油of 油human 油spa3al 油 and 油temporal 油ac3vity 油(Shoval 油& 油Isaacson, 油2007). 油 油 р Visualize 油human 油movement 油on 油2-足D 油plane: 油x-足 油& 油y-足 油axis 油 油longitude 油and 油la3tude; 油z-足axis 油 油3me 油 油 油useful 油for 油analyzing 油sequences 油of 油human 油ac3vity 油(in 油this 油case, 油photo-足taking 油behavior 油of 油 photographers) 油 р Photo-足based 油City 油Modeling: 油 р Understand 油tourist 油site 油aerac3veness 油based 油on 油geotagged 油photos 油(Girardin 油et 油al., 油2008) 油 р Construct 油inter-足city 油travel 油i3neraries 油(De 油Choudhury 油et 油al., 油2010) 油 р Generate 油personalized 油Point-足of-足Interest 油(POI) 油recommenda3ons 油of 油where 油to 油go 油in 油a 油city 油based 油on 油 the 油user's 油travel 油history 油in 油other 油ci3es 油(Clements 油et 油al., 油2010) 油 油 油Approaches 油focus 油on 油describing 油loca3ons, 油not 油on 油鍖ne-足grained 油within-足city 油routes 油that 油connect 油 them 油 р Non-足e鍖ciency 油Driven 油Route 油Planners 油 р Automa3c 油genera3on 油of 油travel 油plans 油based 油on 油millions 油of 油photos 油(Lu 油et 油al., 油2010) 油 р Personalized 油data-足driven 油travel 油route 油recommenda3ons 油(Cheng 油et 油al. 油2011) 油 油 油Systems 油geared 油towards 油recommending 油hotspots 油and 油popular 油routes, 油not 油o鍖-足beat 油explora3on 油 42 routes 油
Sequence Alignment Overview! Input: 油two 油sequences 油over 油the 油same 油alphabet 油 Output: 油an 油alignment 油of 油the 油two 油sequences 油 Example: 油 р 油Source: 油GCGCATGGATTGAGCGA 油 油 油Target: 油 油TGCGCCATTGATGACCA 油 р A 油possible 油alignment: 油 油 油-足GCGC-足ATGGATTGAGCGA 油 油 油TGCGCCATTGAT-足GACC-足A 油 Three 油opera3ons 油(each 油with 油cost): 油 р Perfect 油matches 油(MATCH) 油 р Mismatches 油(DEL) 油 р Inser3ons 油& 油dele3ons 油(INDEL) 油 р The 油less 油distance 油cost, 油the 油higher 油the 油similarity 油between 油two 油sequences 油 (Shoval & Isaacson, 2007) 43
Multiple Sequence Alignment Overview! р Used 油ClustalTXY 油sonware 油(Wilson 油et 油al., 油2008) 油for 油photo 油alignment: 油 р makes 油full 油use 油of 油mul3ple 油pairwise 油sequence 油alignments, 油where 油alignments 油are 油computed 油for 油 similarity 油in 油parallel 油 р uses 油a 油progressive 油heuris3c 油to 油apply 油mul3ple 油sequence 油alignment 油(MSA) 油 р allows 油elements 油to 油be 油represented 油with 油up 油to 油12-足character 油words, 油which 油allows 油unique 油 representa3on 油of 油small 油map 油regions, 油used 油for 油represen3ng 油the 油geotagged 油photos 油 р to 油deal 油with 油di鍖erences 油in 油sequence 油length, 油ClustalTXY 油adds 油gap 油openings 油and 油extensions 油to 油 sequences. 油 油 р MSA 油in 油3 油stages: 油 油1) 油Pairwise 油alignments 油are 油computed 油for 油all 油sequences 油 油2) 油Aligned 油sequences 油are 油grouped 油together 油in 油a 油dendogram 油based 油on 油similarity 油 油3) 油Dendogram 油used 油as 油a 油guide 油for 油mul3ple 油alignment 油 44

More Related Content

Photographer Paths: Sequence Alignment of Geotagged Photos for Exploration-based Route Planning

  • 1. Photographer Paths: Sequence Alignment of Geotagged Photos for Exploration-based Route Planning! Feb. 油26, 油2013 Abdallah 油Abdo 油El 油Ali 油 Sicco 油van 油Sas 油 Frank 油Nack 油 h6p://sta鍖.science.uva.nl/~elali/ 油
  • 2. Outline! I. Introduc3on 油 II. Photographer 油Paths 油 III. User 油Evalua3on 油 IV. Results 油 V. Discussion 油& 油Future 油Work 油 2
  • 4. 4
  • 5. 5
  • 7. 油 油We 油dont 油always 油want 油to 油 supply 油user 油preferences 油 7
  • 8. 油Social 油and 油local 油interpreta3on 油of 油city 油places 油and 油routes 油 8
  • 9. O鍖-足the-足beaten 油track, 油social 油trails, 油 9 油 油Lonely 油Planet! 油 油
  • 10. 油Assump3on: 油 油Loca3ons 油of 油photographs 油are 油poten3ally 油interes3ng 油 10
  • 11. 油But 油sequen3al 油property 油needs 油to 油be 油captured! 油 11
  • 13. 油Analysis 油of 油mobility 油behavior 油of 油city 油 photographers: 油 р where 油photographers 油have 油been 油 р in 油what 油order 油they 油have 油been 油 there 油 р how 油closely 油their 油movements 油 parallel 油those 油of 油other 油 photographers 油 13 By Keiichi Matsuda via supercolossal
  • 14. Research Questions! 油How 油can 油walkable 油route 油plans 油be 油automaCcally 油generated 油for 油residents 油(and 油tourists) 油 that 油would 油like 油to 油explore 油a 油city? 油 油And 油are 油these 油route 油plans 油desirable? 油 油Three 油factors: 油 油1) 油Which 油data 油sources? 油 油2) 油Which 油methods 油to 油generate 油routes? 油 油 油 油 油 油 油 油 油 油 油3) 油User 油percep3ons 油compared 油to 油fastest 油 and 油popular 油routes? 油 油 14
  • 16. Approach! 油1) 油Crawl 油Flickr 油 油geotags, 油3mestamps 油 油2) 油Map 油each 油geotag/loca3on 油in 油a 油sequence 油to 油a 油cell 油in 油a 油par33oned 油grid 油map 油 油3) 油Mul3ple 油Sequence 油Alignment 油on 油photographer 油routes 油to 油鍖nd 油aligned 油loca3on 油sequences 油 油 油These 油alignments 油are 油Photographer 油Route 油Segments 油(PRSs) 油 16
  • 17. Dataset! р Flickr 油geotagged 油photos 油within 油Amsterdam, 油The 油Netherlands 油 油 р Area: 油17.3 油km 油N-足S 油and 油24.7 油km 油E-足W 油(center) 油 р 5-足year 油period 油(Jan. 油2006 油-足 油Dec. 油2010) 油 р Aeributes: 油 油 р owner 油ID 油 р photo 油ID 油 р date 油and 油3me-足stamp 油 р la3tude 油and 油longitude 油(street 油level 油accuracy) 油 油Database: 油426,372 油photos 油 17
  • 18. Preprocessing! р Sequence 油inference 油with 油following 油constraints: 油 р photo 油taken 油within 油4 油hours 油from 油previous 油photo 油and 油in 油same 油order 油 р minimum 油2 油or 油more 油di鍖erent 油loca3ons 油(or 油nodes) 油 р early 油experiments 油determined 油125 油x 油125m 油cells 油in 油center 油of 油Amsterdam 油grid 油suitable 油 油1691 油routes 油(average 油length 油of 油9.92 油loca3ons) 油 油1130 油unique 油photographers 油 18
  • 19. Sequence Alignment! р To 油鍖nd 油photographer 油paths 油from 油Photographer 油Route 油Segments 油(PRSs), 油constraints 油set: 油 р PRS 油has 油minimum 油4 油photographers 油with 油minimum 油2 油aligned 油nodes/loca3ons 油 !#**% Photographer Route Segments (PRSs)! !"#$% !"**% !+**% %&'()*+$,*-.*)/*0$ !***% $**% ##&% +%,-./.012,-314% 油 油 油 油231 油PRSs 油 油 油 油 #**% '!&% )%,-./.012,-314% 油(average 油length 油of 油2.61 油nodes) 油 "%,-./.012,-314% "**% !"#$ +**% !'(% !*!% +'% !*% !% )'% *% +% )% "% '% 1)'-.*$&2/342)0$ 19
  • 21. PRS Aggregation! р Modi鍖ed 油Dijkstras 油shortest 油path 油algorithm 油 & P1N1 P2N1 & # ! P1N2 ! " # !" #$% Start P1N3 P2N2 End 21
  • 22. PRS Aggregation to Crude Routes! PRSs CM Photographer Route WW Photographer Route 22
  • 24. Laboratory Study Design! р ~45 油min. 油Quan3ta3ve/Qualita3ve 油lab-足based 油study 油 р 15 油par3cipants 油(10 油m, 油5 油f) 油aged 油between 油21-足35 油(M 油= 油29.2; 油SD 油= 油3.3) 油 р Interac3ve 油web-足based 油prototype 油route 油planner 油 р Expert 油route 油evalua3on 油by 油city 油residents 油(lived 油in 油Amsterdam 油> 油1 油year) 油 р Plain 油routes 油to 油avoid 油informa3on 油type 油bias 油 24
  • 25. Laboratory Study Design! р Two 油scenarios: 油 р Route 油1: 油Central 油Sta3on 油to 油Museumplein 油 油anernoon 油scenario 油favoring 油explora3on 油 р Route 油2: 油Waterlooplein 油to 油Westerkerk 油 油evening 油scenario 油favoring 油e鍖ciency 油 р Baseline 油comparisons: 油 油 р Photo 油Density 油(PD) 油route: 油highest 油density 油of 油photos 油(over 油5 油year 油period) 油in 油grid 油cells 油 along 油route 油 油 р Google 油Maps 油(GM) 油route: 油shortest 油route 油between 油two 油loca3ons 油 р Counterbalanced 油within-足subject 油design 油 油 р Route 油Varia3on 油(IV): 油Photographer 油Paths 油vs. 油Photo 油Density 油vs. 油Google 油Maps 油 25
  • 26. Central Station to Museumplein (CM)! Photographer Paths Photo Density Google Maps route (5.36 km) route (3.83 km) route (3.35 km) 26
  • 27. Waterlooplein to Westerkerk (CM)! Photographer Paths route (2.28 km) Photo Density route (2.60 km) Google Maps route (1.59 km) 27
  • 28. Laboratory Study Design! 油Data 油collected: 油 1. AerakDi鍖2 油(Hassenzahl, 油2003) 油UX 油ques3onnaire 油responses 油[7-足point 油seman3c 油 di鍖eren3al 油scale]: 油Usability, 油Hedonic 油Quali3es 油(Iden3ty, 油S3mula3on), 油Aerack3veness 油 2. Two-足part 油semi-足structured 油interviews 油 油Part 油1: 油Route 油preferences, 油feedback 油on 油Photographer 油Paths 油 油Part 油2: 油Inves3ga3on 油of 油visualized 油informa3on 油types 油(visualized 油info 油type 油handouts): 油 油 油a) 油Google 油maps 油 油 油b) 油Color 油coded 油PRSs 油(PP 油route) 油 油 油c) 油Density 油geopoints 油(PD 油route) 油 油d) 油Thumbnail 油photo 油geopoints 油 油 油e) 油Foursquare 油POIs 油 28
  • 29. a) d) b) e) c) 29
  • 30. Web Survey Study! р Short 油web-足based 油survey 油for 油CM 油and 油WW 油routes 油and 油varia3ons 油 р Basic 油demographics 油collected: 油age, 油gender, 油years 油in 油Amsterdam 油 р Sta3c 油route 油images, 油no 油counterbalancing 油 р 82 油par3cipants 油(55 油m, 油27 油f) 油aged 油between 油17-足62 油(M= 油27.6; 油SD= 油6.1) 油 油 р Most 油lived 油in 油Amsterdam 油for 油more 油than 油3 油years 油(44/82) 油 油 р Some 油between 油1-足3 油years 油(15/82) 油 р Less 油than 油a 油year 油(11/82) 油 油 р Only 油visited 油before 油(12/82) 油 30
  • 32. AttrakDiff2 ! Central Station to Museumplein (CM) Route "# ** ** ** ** * ** ** * $# %# !"#$% &# !%# !$# !"# '()*+),-#./)0123#4'.5# 6789:1-#./)0123#!#;87:,23# 6789:1-#./)0123#!# =>()-,?7:7@@#4=AA5# 46.!;5# <,+/0),9:#46.!<5# &'("$)'*$% ** !"#"$%$& * !"#"$%$' 'B929*()CB7(#')2B@# 'B929#D7:@123# E99*07#F)C@# 32
  • 33. AttrakDiff2 ! Waterlooplein to Westerkerk (WW) Route ** ** "# ** ** $# %# !"#$% &# !%# !$# !"# '()*+),-#./)0123#4'.5# 6789:1-#./)0123#!#;87:,23# 6789:1-#./)0123#!# =>()-,?7:7@@#4=AA5# 46.!;5# <,+/0),9:#46.!<5# &'("$)'*$% !"#"$%$& ** 'B929*()CB7(#')2B@# 'B929#D7:@123# E99*07#F)C@# 33
  • 34. Route Preference! Lab 油Study 油 р CM 油route: 油most 油chose 油to 油follow 油the 油PP 油route 油(9/15), 油PD 油route 油(4/15), 油GM 油route 油(2/15) 油 р One 油of 油the 油routes 油[PP] 油was 油long 油and 油took 油many 油detours, 油and 油I 油thought 油that 油was 油a 油 very 油aFracHve 油route! 油 р WW 油route: 油most 油chose 油to 油follow 油GM 油route 油(10/15), 油PD 油route 油(4/15), 油no 油route 油(1/15) 油 油 р You 油are 油going 油for 油co鍖ee 油so 油you 油just 油want 油to 油get 油there, 油unlike 油in 油the 油鍖rst 油[CM] 油 scenario 油where 油it 油is 油a 油nice 油day 油and 油you 油have 油Hme. 油 Web 油Survey 油 ≒ 油 油 油 油 油CM 油route: 油GM 油(40/82), 油PD 油(23/82), 油PP 油route 油(10/82), 油neither 油(9/82) 油 ≒ 油No 油experimenter 油steering; 油many 油Amsterdam 油residents 油know 油the 油city 油already 油quite 油 well! 油 ≒ 油I 油would 油not 油easily 油walk 油these 油routes... 油who 油in 油Amsterdam 油walks? 油;) 油 ≒ 油 油 油 油 油WW 油route: 油GM 油(67/82), 油PD 油(6/82), 油PP 油route 油(3/82), 油neither 油(6/82) 油 34
  • 35. Digital Information Aids! Lab 油Study 油 油 油 油 油 油 油 油Interview: 油Part 油1 油 р How 油many 油persons 油(focus 油on 油city 油photographers) 油took 油a 油given 油route 油segment 油over 油a 油certain 油3me 油 period 油(e.g., 油1 油year)? 油 р Useful 油(8/15) 油for 油exploring 油a 油city 油one 油already 油knows 油 р Not 油sure 油(4/15) 油 р Depends 油on 油which 油photographers 油(2/15) 油 р Not 油for 油me 油(1/15) 油 油Interview: 油Part 油2 油 р Found 油PP 油info 油type 油aerac3ve 油(10/15), 油but 油combine 油with 油Photo 油thumbnails 油(3/10) 油and 油POIs 油(3/10) 油 35
  • 36. Digital Information Aids! Web 油Survey 油 р POIs 油along 油a 油route 油(51x) 油 р Route 油distance 油(51x) 油 р Comments 油along 油a 油route 油(ranked 油by 油highest 油ra3ngs 油or 油recency) 油(24x) 油 р Expert 油travel 油guides 油(22x) 油 р Photos 油of 油route 油segments 油(17x) 油 р No 油digital 油aids 油(13x) 油 р Number 油of 油photographers 油that 油took 油a 油given 油path 油over 油a 油Hme 油period 油(9x) 油 р Number 油of 油photos 油along 油a 油route 油over 油a 油3me 油period 油(9x) 油 36
  • 38. Discussion! р Discrepancy 油between 油lab-足study 油and 油web 油survey 油 р Quick 油web 油survey 油insu鍖cient? 油 油 р Visualiza3on/explana3on 油of 油digital 油aids 油important? 油 р Proof-足of-足concept 油approach 油requires 油real-足world 油outdoor 油evalua3on 油 р Di鍖erent 油street 油grid 油network 油 р Scalability 油to 油larger 油ci3es 油 р More 油context-足awareness 油 38
  • 39. Take Home Message! Going 油towards 油data-足driven 油explora3on-足based 油route 油planners 油 р Some3mes 油its 油the 油journey, 油not 油the 油des3na3on 油 р A 油quan3ta3ve 油approach 油may 油oversimplify 油human 油needs 油for 油explora3on 油 р But 油some3mes 油we 油want 油an 油automa3c 油solu3on, 油so 油as 油not 油to 油be 油bothered 油 with 油supplying 油user 油preferences 油and 油encounter 油serendipity 油 39
  • 40. Questions 40 h6p://sta鍖.science.uva.nl/~elali/ 油
  • 41. References! 油1. 油Cheng, 油A.-足J., 油Chen, 油Y.-足Y., 油Huang, 油Y.-足T., 油Hsu, 油W. 油H., 油and 油Liao, 油H.-足Y. 油M. 油Personalized 油travel 油 recommenda3on 油by 油mining 油people 油aeributes 油from 油community-足contributed 油photos. 油In 油Proc. 油 MM 油11, 油ACM 油(2011), 油8392. 油 油2. 油M. 油Clements, 油P. 油Serdyukov, 油A. 油P. 油de 油Vries, 油and 油M. 油J. 油Reinders. 油Using 油鍖ickr 油geotags 油to 油predict 油 user 油travel 油behaviour. 油In 油Proc. 油SIGIR 油10, 油pages 油851852. 油ACM 油Press, 油2010. 油 3. 油M. 油De 油Choudhury, 油M. 油Feldman, 油S. 油Amer-足Yahia, 油N. 油Golbandi, 油R. 油Lempel, 油and 油C. 油Yu. 油 Automa3c 油construc3on 油of 油travel 油i3neraries 油using 油social 油breadcrumbs. 油In 油Proc. 油HT 油10, 油pages 油 3544. 油ACM 油Press, 油2010. 油 4. 油F. 油Girardin, 油F. 油Calabrese, 油F. 油D. 油Fiore, 油C. 油Ra|, 油and 油J. 油Blat. 油Digital 油footprin3ng: 油Uncovering 油 tourists 油with 油user-足generated 油content. 油IEEE 油Pervasive 油Compu3ng, 油7:3643, 油October 油2008. 油 5. 油N. 油Shoval 油and 油M. 油Isaacson. 油Sequence 油alignment 油as 油a 油method 油for 油human 油ac3vity 油analysis 油in 油 space 油and 油3me. 油Annals 油of 油the 油Associa3on 油of 油American 油Geographers, 油97(2):282297, 油2007. 油 6. 油A. 油Vaccari, 油F. 油Calabrese, 油B. 油Liu, 油and 油C. 油Ra|. 油Towards 油the 油socioscope: 油an 油informa3on 油system 油 for 油the 油study 油of 油social 油dynamics 油through 油digital 油traces. 油In 油Proc. 油GIS 油09, 油pages 油5261. 油ACM 油 Press, 油2009. 油 7. 油Hassenzahl, 油M., 油Burmester, 油M., 油and 油Koller, 油F. 油AerakDi鍖: 油Ein 油Fragebogen 油zur 油Messung 油 wahrgenommener 油hedonischer 油und 油pragma3scher 油Qualit即at. 油Mensch 油& 油Computer 油2003. 油 Interak3on 油in 油Bewegung 油(2003), 油187196. 油 8. 油Lu, 油X., 油Wang, 油C., 油Yang, 油J.-足M., 油Pang, 油Y., 油and 油Zhang, 油L. 油Photo2trip: 油genera3ng 油travel 油routes 油 from 油geo-足tagged 油photos 油for 油trip 油planning. 油In 油MM 油10, 油ACM 油(2010), 油143152. 油 9. 油Wilson, 油C. 油Ac3vity 油paeerns 油in 油space 油and 油3me: 油calcula3ng 油representa3ve 油hagerstrand 油 trajectories. 油TransportaHon 油35 油(2008), 油485499. 油 41
  • 42. Related Work! р Sequence 油Alignment 油(SA) 油methods: 油 р Borrowed 油from 油bioinforma3cs 油and 油later 油3me 油geography 油 р Time 油geography 油systema3cally 油analyzes 油and 油explores 油the 油sequen3al 油dimension 油of 油human 油spa3al 油 and 油temporal 油ac3vity 油(Shoval 油& 油Isaacson, 油2007). 油 油 р Visualize 油human 油movement 油on 油2-足D 油plane: 油x-足 油& 油y-足 油axis 油 油longitude 油and 油la3tude; 油z-足axis 油 油3me 油 油 油useful 油for 油analyzing 油sequences 油of 油human 油ac3vity 油(in 油this 油case, 油photo-足taking 油behavior 油of 油 photographers) 油 р Photo-足based 油City 油Modeling: 油 р Understand 油tourist 油site 油aerac3veness 油based 油on 油geotagged 油photos 油(Girardin 油et 油al., 油2008) 油 р Construct 油inter-足city 油travel 油i3neraries 油(De 油Choudhury 油et 油al., 油2010) 油 р Generate 油personalized 油Point-足of-足Interest 油(POI) 油recommenda3ons 油of 油where 油to 油go 油in 油a 油city 油based 油on 油 the 油user's 油travel 油history 油in 油other 油ci3es 油(Clements 油et 油al., 油2010) 油 油 油Approaches 油focus 油on 油describing 油loca3ons, 油not 油on 油鍖ne-足grained 油within-足city 油routes 油that 油connect 油 them 油 р Non-足e鍖ciency 油Driven 油Route 油Planners 油 р Automa3c 油genera3on 油of 油travel 油plans 油based 油on 油millions 油of 油photos 油(Lu 油et 油al., 油2010) 油 р Personalized 油data-足driven 油travel 油route 油recommenda3ons 油(Cheng 油et 油al. 油2011) 油 油 油Systems 油geared 油towards 油recommending 油hotspots 油and 油popular 油routes, 油not 油o鍖-足beat 油explora3on 油 42 routes 油
  • 43. Sequence Alignment Overview! Input: 油two 油sequences 油over 油the 油same 油alphabet 油 Output: 油an 油alignment 油of 油the 油two 油sequences 油 Example: 油 р 油Source: 油GCGCATGGATTGAGCGA 油 油 油Target: 油 油TGCGCCATTGATGACCA 油 р A 油possible 油alignment: 油 油 油-足GCGC-足ATGGATTGAGCGA 油 油 油TGCGCCATTGAT-足GACC-足A 油 Three 油opera3ons 油(each 油with 油cost): 油 р Perfect 油matches 油(MATCH) 油 р Mismatches 油(DEL) 油 р Inser3ons 油& 油dele3ons 油(INDEL) 油 р The 油less 油distance 油cost, 油the 油higher 油the 油similarity 油between 油two 油sequences 油 (Shoval & Isaacson, 2007) 43
  • 44. Multiple Sequence Alignment Overview! р Used 油ClustalTXY 油sonware 油(Wilson 油et 油al., 油2008) 油for 油photo 油alignment: 油 р makes 油full 油use 油of 油mul3ple 油pairwise 油sequence 油alignments, 油where 油alignments 油are 油computed 油for 油 similarity 油in 油parallel 油 р uses 油a 油progressive 油heuris3c 油to 油apply 油mul3ple 油sequence 油alignment 油(MSA) 油 р allows 油elements 油to 油be 油represented 油with 油up 油to 油12-足character 油words, 油which 油allows 油unique 油 representa3on 油of 油small 油map 油regions, 油used 油for 油represen3ng 油the 油geotagged 油photos 油 р to 油deal 油with 油di鍖erences 油in 油sequence 油length, 油ClustalTXY 油adds 油gap 油openings 油and 油extensions 油to 油 sequences. 油 油 р MSA 油in 油3 油stages: 油 油1) 油Pairwise 油alignments 油are 油computed 油for 油all 油sequences 油 油2) 油Aligned 油sequences 油are 油grouped 油together 油in 油a 油dendogram 油based 油on 油similarity 油 油3) 油Dendogram 油used 油as 油a 油guide 油for 油mul3ple 油alignment 油 44