Docume
D5.1 Inte
Project N
FP7‐201
Contract
M10
Docume
Stefan Fo Reza Raw Gerd Ko Wolfgan
Marcus H
Abstract
This del aware u concept informat propose interface transpor discuss d data flow our mod the raw level and
ent Title
ent‐aware U
Number
1‐7‐287661
t Delivery Da
ent Author o
oell
wassizadeh rtuem ng Apolinarsk
Handte
t
iverable is t user interface
for improve tion about u
in this docu e layers whic rt layer, per details of the ws and pred del is the clo
context sen d tangible kn
ser Interface
Project A
GAMBAS
ate Actual
Novem
r Reviewer
ki
the first acti es which are ed user intera users’ curren ument a com ch cover diffe sonal travel e user‐interfa dictions of fu ose relationsh sed by the a nowledge for
e Model 1
Acronym
Delivery Da
mber 30th, 20
ivity in WP5 e part of th action based nt and future mprehensive
erent aspect behaviour a ace system b uture travel
hip of interfa adaptive data r the user rel
Project Titl
Generic Ad Autonomo
ate Deliv
012 R (Re
5 and descri e GAMBAS d on user int e informatio
user interfa ts of mobility and social t based on nov
behaviour a ace design a a acquisition levant to his
le
daptive Midd us Services
verable Type
eport)
Organiza
OU OU OU UDE UDE
bes the me system. Inte erfaces whic on needs. In ace model w y in urban sc ravel behavi vel user inte and context and data pro n framework
current and
dleware for B
e Del
PU
ation
thodology fo ent‐awarene ch are augme
order to ad which is base enarios: tran iour. For eac rface concep parameters. ocessing algo presented i
future situa
Document V
1.0
Behavior‐driv
liverable Acc
(Public)
Work Packa
WP5 WP5 WP5 WP5 WP5
or designing ess refers to ented with s ddress this is ed on a stack nsit layer, qu ch of the la pts, design s . The key no orithms to tr
in D2.1.1 int ations.
Version
ven
cess
age
g intent‐ a novel semantic ssue, we k of four uality‐of‐
0.1 OU Outline of document structure 0.2 OU Description of user interface model 0.3 OU Description of prediction system 0.4 OU Description of related work 0.5 OU Description of system architecture 0.6 OU Description of requirement coverage 0.7 OU Introduction and conclusion
0.8 OU Description of abstract
0.9 UDE Reviewed the whole document
1.0 UDE Final review for submission
Table
of
Conte
roduction ... Purpose .... Scope ... Structure .. er Interface I The Madrid .1 Digita .2 Trans Survey of C Research T .1 Perso .2 Group
Conclusion
ent‐Aware U Transport .1 Motiv
Personal B .1 Motiv .2 Novel .3 User I .4 User I .5 Data a
Social Beha .1 Motiv
ssues Relate d Public Tran al Services ....
port Data .... Commercial Topics ... onal Behaviou
p Behaviour n ... ser Interface Layer ... vation and O Interface Ser Interface Ske and Algorith
‐Transport La l Concepts ... Interface Ser Interface Ske and Algorith Behaviour Lay
vation and O l Concepts ... Interface Ser Interface Ske and Algorith aviour Layer vation and O nsport System
...
...
Mobile Publ ...
bjectives ... rvices ... bjectives ... ...
3.4 tem Architec User Interf .1 Prese Personal B .1 Input .2 Outpu .3 Predic Social Beha .1 Input .2 Predic .3 Predic quirements C Intent‐rela Privacy‐rel nclusion ... liography ...
Interface Ser Interface Ske and Algorith cture ... face System ntation ... roller ... Inference .... Access ... rience Loggin
cy ... n into GAMB Algorithms .. Transportati
...
ut ... ction Algorit Behaviour: Da
...
ut ... ction Algorit aviour: Data Data ... ctions ... ction Algorit Coverage An ated Require ated Require ...
BAS architect ...
nd Predictio ...
...
...
List
of
Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 1 Figure 12 Figure 13 Figure 14 Figure 15 Figure 16 Figure 17 Figure 18 Figure 19
f
Figures
– GAMBAS M – Abstract R – User Inter – EMT Andr – EMT iPhon – Madrid Pu – Layered U – Typical vis – Simple me 0 – An exam 1 – An exam 2 – Design Sk 3 – Design Sk 4 – Envisione 5 – Envisione 6 – Envisione 7 – Architect 8 – User Inte 9 – Integrati
Middleware Representatio rface of EMT roid Applicat
ne Applicatio ublic Transpo User Interface sualization of etro map visu ple of how a ple of how a ketch for Cog ketch for Cog ed design of ed design of ed design of ture of the U erface System
on of User‐In
Architecture on of Madrid Transport W ion ... on ... ortation Syst e Model ... f the transpo ualization on a QoT map co a QoT map co gnitive Map gnitive Map a Social Tran a Social Tran a Social Tran User Interface m Data Flow
nterface Syst
e with User I d Public Tran Web Applicat ... ... tem ... ... ortation laye n smartphon ould be reali ould be reali 1 ... 2 ... nsport Map nsport Map nsport Map e System ... Diagram ... tem into GA
nterface Com nsport System tion ...
... ... ... ... er ... e ... sed. ... sed. ... ... ... 1. ... 2 ... 3 ... ... ... MBAS Archit
mponent ... m. ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... tecture ...
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
... 2
... 4
... 5
... 5
... 6
... 6
... 10
... 12
... 12
... 14
... 14
... 16
... 16
... 18
... 18
... 19
... 20
... 21
List
of
Table 1:
f
Tables
Feature Commparison bettween Transport Apps ...
1
Int
The GAM of smart realised provide services context. GAMBAS systems user int behaviou capable doing so
MBAS projec t cities supp by developi adaptive an for sensing,
Work packa S application em uses sem commendatio
urs of intere rtation and t
urpose
verable desc erfaces. For t encompass erface mod ural mobility
of displayin o, our work g nostic and ge lized views o cenarios.
shows the In objectives a Developmen information approach to achieve the Developmen information transport ne Developmen the notion o travellers ac Developmen travel behav transport sys
on
ct envisions porting key ng a generic d predictive , processing, age 5 is conc ns and the d mantic infor ons. As the
st relates to ravel choice
cribes the ba this purpose ing dimensio
el lies in th y patterns f g and augm goes beyond eneral inform
on the trans
ntent‐Aware nd contribut nt of a conc
needs for t exploit the intent‐aware nt of a new
beyond tim etwork nt of a new c of cognitive
tual behavio nt of a socia
mation abou e applicatio
the use of p s.
asic theoretic , we explore ons of perso he combinat from travel
enting this i
existing mo mation. The e
portation sy
e User Interfa tions of this d ceptual user the design o relationship e user interfa
interface co me and locati
concept for t maps to ena our in reality
al interface ghlight socia
of behaviou urban life s f Things mid n to people [ ting informa the develop ent‐aware u ut user beh on domain
public transp
c model for e a novel des onal and soc tion of tech
histories an information obile travel g exploitation ystem is a ne
ace Service i deliverable a r interaction of mobile tra between be ace concept oncept for ion to disco the persona able travel e
concepts w al relations a
ural‐driven a such as pub
dleware and [D1.3.1]. The ation about pment of a g user interface aviours to p
of GAMBAS portation and
the design a ew approach
n the contex are:
n model wh avel guides. ehavioural co making use ver and high lization of pu experiences w
hich gives in among friend
pplications f blic transpor d intelligent e middlewar how people generic user es. Intent‐aw provide perso S is public d user interfa
nd developm r novel mob ehaviour. Th extract user er interface e mobility n rip planners utines of use for improve
xt of the GAM
hich covers The model ontext data a of a broade hlight crowd ublic transpo which are cl nsight into s ds and like‐m
for future ge rtation. This cloud servic re consists o e behave in
interface se wareness me onalized info transportat concepts w eeds of trav which mostl ers for the cr ed user inter
MBAS Middle
different as is based on and user inte er notion of dedness in th ort systems losely couple social inform minded peop
eneration vision is ormation tion, the relate to
In summ Model (S mining m discover satisfact
1.2
Sc
This doc concepts describe GAMBAS describe
With res consorti with me explores transpor
1.3
St
This deli mobile t readily a
Fig
mary, the key Section 4), d methods for r routines in tion with pub
cope
cument is t s and techni ed in this do S application ed in D5.3.
spect to the um ([D1.3.1 eans to acq s opportunit rtation and b
tructure
iverable is st travel system available on
gure 1 – GAMB
y contributio design sketc r understand n travel beha blic transport
the first del iques for int ocument wil n prototype (
e overall GA ], [D2.1.1]) w uire real‐tim ties of how build intent‐a
tructured as ms. As part the market,
Servic
A Inte
1
2
3
V
BAS Middlewar
ns of this de hes of user ding and pr aviour), as w
t systems.
liverable in tent‐aware u
l be used in (D5.2). The r
MBAS system with a focus me context
to make e aware user in
follows. In of this anal
as well as p
ce
Discove
M
Adaptive Data Acquisiton ent‐aware Us
Interfaces Prot
Validates
re Architecture
eliverable are interfaces th redicting use well as user
WP5 and i user interfac n the develo result of use
m, the deliv s on human data of tra effectively u nterfaces.
Section 2 w lysis, we dis present an o G
ery
and
Co
Middlewar
a ser
totype Applic
with User Inte
e: the definit hat realise t er behaviou experience
is concerned ces. The conc opment of c r studies to e
verable exte aspects. Wh avellers on m use of this d
we provide a scuss existing overview of c GAMBA
GAMBAS M
GAMBAS Com
ommunic
re
Aut
o
m
a
ted
Priv
acy
‐
P
res
er
a
tio
n
cation
5
Enable
rface Compone
tion of an ab his model (S r (intent pre
metrics tha
d with deve cepts, metho concrete use evaluate use
nds previous hile both D1 mobile devi data to eng
review of th g mobile ap current rese
AS Approach
Middleware mponents
ation
In
te
ro
p
e
ra
b
le
Dat
a
Mod
e
lli
ng
an
d
P
roce
ssi
n
g
4
es
ent
bstract User Section 4), b ediction me at capture th
eloping fund ods and des er interfaces er experience
s deliverable 1.3.1 and D2
ices, this de gage users
he state‐of‐t pplications w arch in mob
Interface behaviour ethods to he users’
damental ign ideas s for the es will be
es of the 2.1.1 deal eliverable in public
guides a model a layers of its integ intent p relevant phase o summar
nd trip advis nd present t f the architec gration into t rediction co t to the use f the projec y in Section
sors. In Secti the design ap
cture. In Sec the overall G omponent w
r interface s ct, is discuss
7.
ion 3, we int pproach inclu
tion 4, we gi GAMBAS arc hich is part system, whi sed in Sectio
troduce the uding motiva ive an overvi chitecture. S of our inter ch have bee on 6. Finally
conceptual a ation, object iew of the us Subsequently rface system en defined i y, we conclu
architecture ives and con ser interface y, we introd m. The cover in the requi de this deliv
of the user ncepts for ea e system and duce in Secti rage of requ irement spe verable with
interface ach of the
2
Us
The GAM the pub worked develope describe applicati
2.1
T
The com km2, and Madrid. is clear largest m from a h In additi the who lines ope 15 minu which ru a repres
1
http://w
ser
Interf
MBAS middle lic transport out (and w ed for the ci e the Madrid
ions.
The Madri
mmunity of M d comprises
Thus large n demand for metro railwa high level of c
on to the M ole city and is
erate everyd utes, depend un along 27 d
entation of t
www.citymayo
face
Issue
eware and s tation doma will be publi ity of Madrid d public tran
d Public T
Madrid is loc a population numbers of p
r sophisticat ay in the wo
congestion d etro system s run by GAM day between ding on the
different rou the central a
Figure 2 – Abs
ors.com/trans
es
Relate
ervices will in. Although shed as D6 d, facilitated sport scenar
Transport
cated in the n of about 5 people from t
ted transpor rld, with 13 during peak h
s, Madrid ha MBAS projec n 6 am and 1 time of day utes. In tota nd busiest p
stract Represen
sport/madrid‐
ed
to
Pub
be validated h the exact
.1.1) it is c d by project rio and analy
rt System
e centre of S
million peo the surround rt informatio metro lines hours [30]. as an extensi ct partner Em 11.30 pm, w y. There are al the EMT fl part of the M
ntation of Mad
‐metro.html
blic
Trans
d in the cont details of th lear that th partners EM yse user inte
Spain, covers ple, with 3 m ding area co on systems.
and 296 me
ive city bus n mpresa Mun with buses lea
also night b eet comprise Madrid public
drid Public Tran
sportatio
ext of a pro he validation
e prototype MT and ETRA erface issues
s an area of million peop
mmute on a Although M etro station1
network, wh icipal de Tra aving at inte buses, know es 2,022 veh transport sy
nsport System.
on
System
ototype appli n plan are s e application A. In the follo s of mobile t
f approximat le living in th
daily basis a Madrid has
1
, the city is
hich covers p ansporte (EM ervals betwe wn as “búhos
hicles. Figure ystem.
ms
ication in still being n will be owing we transport
tely 8000 he city of and there the sixth suffering
2.1.1 D
A rich di the next accessed users to (ETA) for
EMT also services
A
w
DigitalServ
igital ecosyst t bus is due, d through a w
find bus rou r all bus stop
o provides a (Figure 4 an A map of the location‐awa real‐time jou waiting time
vices
tem of data the GAMBA web interfac utes, get sche ps.
Figure 3
n official iPh d Figure 5). e Madrid com are station fi urney calcula e and distanc
and service AS partners E
ce and a mo eduled depa
3 – User Interfa
one and And Both mobile mmuter netw
nder, ator, and ce calculator
Figure 4 –
s exists for t EMT and ETR
bile phone. T arture and ar
ace of EMT Tran
droid applica e applications work (metro
.
EMT Android A
the Madrid t RA have set The web inte rrival times a
nsport Web Ap
ation with ac s provide acc and bus line
Application
transport sys up an online erface (show and get estim
plication
ccess to the m cess to: es),
stem. To kno e service tha wn in Figure mated times
most freque
ow when at can be 3) allows of arrival
Third‐pa For exam providin lines.
2.1.2 T
An incre three ty network (estimat following
2
https://
3 https:// 4
http://w
arty mobile a mple, the ‚Bu
g personalis
Transport
easingly rich ypes of infor k data (i.e. sto
ted time of a g the Googl
/play.google.c /developers.go www.gtfs‐data
applications us Madrid’2 sation featur
Fi (generated
Data
set of data i rmation are ops and rout arrival = ETA le supported
om/store/app oogle.com/tra a‐exchange.co
Figure 5 –
with extend application f res that allo
gure 6 – Madri from open tra
is available f
available fo tes), (II) stati A). Route da d General T
ps/details?id= ansit/gtfs/refe om/agency/m
– EMT iPhone A
ed functiona for Android ow users to
id Public Trans nsport data by
for public tra or the Mad ic time‐table ta and sche Transit Feed
=es.android.bu erence madrid/
Application
ality are ava phones goes define their
portation Syste y http://openbu
ansport syste rid public tr e data, (III) dy
dules for M Specificatio
usmadrid.apk
ilable for mo s beyond the r preferred b
em usmap.org/)
ems around ransport sys ynamic real‐t
adrid are av on (GTFS) sp
&feature=rela
ost mobile p e EMT applic bus stations
the world. C stem: (I) sta
time bus info vailable as o pecification3
ated_apps
latforms. cation by and bus
Currently tic route ormation pen data
4
XML5. O
Google Now
SeoulBus
de. Journey
with some , there are rtation habit e users to co r to clarify fe applications. h project or d
pplications t y of the app
App
Resource
https://www .google.com/ landing/now
https://play. google.com/s tore/apps/de tails?id=com. astroframe.s eoulbus
https://play. ualisation of
Commerc
phone journ
planer app download fre
that system lication featu
Static or
static
etmap.org/wi org/
cker, London B Automobiles
developers a Map6.
f the Madrid
ial Mobile
ney planning ps mostly re e map syste pplications t rage behavio fuel while dr existing publ for selecting equency in t
atically mine ures is prese
and compan
public trans
e Public T
g application ely on static
em provide that use G our changes. riving8 9.
ic transport g application he Android m
_ Suppo
d, bas
_ curre
locat
_
_
s_in_Madrid
TubeMap Pro
nies to creat
sport system
Transport
ns exist for c informatio
by Google) PS and acc . For instanc
apps, we a ns for the su market. Exce aviour to pro
e 1.
ext Personali
orte location s hom
ent
ion
Only rou based current lo
_
te dedicated
generated f
t Applicat
Madrid and n, such as p
ovide person
isation U
Visual ocation
Inter public‐transf time inform s to monito e application
ubset of all ed importan le Now there nalised serv
I & lization aphor
Sh
ractive h map
ractive
h map pport
a static ap is ported
mapping
available
her cities fer time‐ mation. In
or users’ ns, which
available ce of the titude
_
SG Buses
Madrid Metro|B us|Cerca nias
NYCMat e (Bus & Subway)
Live London Bus Tracker
phones based a persona causes p friendly persona sgbuses
madrid
https://play. bwaymap
https://play. google.com/s tore/apps/de tails?id=com. appeffectsuk. bustracker
https://play.
Research
T
ed interactive categories: (I
group behav
PersonalB
ollowing, ap al sense, the n the collect , GreenSaw ble behavio to collect us activity class l ambient di personal aw behaviour. J l behaviour.
plications th se mobile ap ed contextua w[6] and Ub
our. PEIR (Pe sers’ locatio sification, de isplay on mo
areness abo Jigsaw [6] is
applications that aims at
hat mine and pplications s al informatio iGreen[2] ar ersonal Env on changes, etermines u obile phones out transport
another ap
_ Curre
locat
x Curre
locat
x _
x Curre
locat
x Curre
locat
mparison betw
are now be ironment Im and by usin users’ transp s to give fee
tation activi plication tha
ent ion
Only rou based current lo
ent ion
Only rou based current lo
_
ent ion
Only rou based current lo
ent ion
Routing on curr
ween Transport
eing investiga behaviour m
rsonal beha ontextual info onal usage. applications mpact Facto
g an accele portation m
dback about ty and reinf at provides a
uting on ocation
Inter with
uting on ocation
_ ocation
Inter odification a
viour will be ormation an
that motiva r) [1] uses rometer sen mode. UbiGr t transportat forces user c a user interf
ractive h map
_
oth ractive map only
ractive
ractive
eral research and (II) resea
e described. nd perform r
ate users to GPS data o nsor data an reen [2] pro
tion behavio commitmen reasoning
Another namely informat
2.3.2 G
Another combine PROCAB behaviou recomm users’ be Tulusan through
2.4
C
In sum, transpor not mak location the follo
outes based e. Garcia et
zation by id heckr[15] pe ative multim direct anno tion.
GroupBeha
category o es group beh B [6] uses GP
ur. It tries to end better ehaviour, wh
et al. [16] u providing us
onclusion
there are rt behaviour ke use of per
and proxim owing areas: Incorporatin quality of av User interfa profiles User interfa scenarios (e. reas of inno tion of Work
ocus, in this a s’ usage, suc
contains tra d on user sel al. [11] prop dentifying P ersonalizes modal annot
otation of g
aviour
of applicatio haviour mode PS traces for o incorporate
routes to ta hich includes
use the EcoG sers with eco
n
no applicat . Furthermo rsonalisation
ity. This pro
ng novel con vailable trans ace persona
aces that p .g. tomorrow ovation dire .
area, is “aut h as [10, 11] nsit network lection and a pose a system
OI (point o the routing ation of geo geographical
ons focuses els with adva ovides concre
text parame sport options lisation and rovide user w, tonight) so ectly relate t
omatic trave ]. PECITAS [1 ographical d
data by us
on interac anced user in
s to underst avioural atti Moreover, it eferences an
phones app o reduce fue
rovide adva rface approa
on and use ete evidence
eters into tra s)
adaptation
rs with data orms recomm
of tourists’ of mobility data. The an sers and ac
ctive service nterfaces. tand their ro
anced intera aches of exis of context p e that there
ansport appl
based on
a and infor effective tra arch objecti
Most of these a route reco formation of e a ranked li mendation, r routes with impaired p notation is cquisition of
es for group oute prefere route recom easoning abo corporate d
ctive service sting or prop parameters is
is ample spa
lications (for individual a rmation abo ansport plann
ves of WP5
e projects ha ommendatio
f Bolzano, It ist routes in route genera hin public t
pedestrians based on tw f directly ob
ps of travel ences and le mmendation out context‐ driver behav
es by minin posed applica s primarily l ransport. through wo parts, bservable
llers and earn their and thus
‐sensitive
3
In
In this s The mo particula and not In that s GAMBAS
Each lay design a
‐ ‐ ‐ A From bo
‐ T f
‐ T e e f
‐ T
ntent
‐
Awa
ection we d del consists ar user inter application s sense the us
S.
er adds a pa spects of (so Data (about User Interfac Algorithms ( ottom to top The transpo forms the ba route finding The Quality experience o entails elem finding. The Persona people’s mo
are
User
describe a ne s of four lay rface design
specific, the ser interface
articular desi ome or) all fo
transportati ce elements for generati
these layers rt layer (TL) aseline for a g and mobile
of Transpor of a public ments that h
al Behaviou obility pattern
Interfac
ew user inte yers (as rep aspect. Whi model is tai model refle
Figure 7 – La
ign aspect; t our layers. Ea
ion and user (i.e. widgets ng user inter s introduce th
focuses on ll layers abo e transport a t layer (QoT transport sy help users m
r Layer (PB ns, in particu
e
Model
erface mode presented in ile the unde
lored for do ects publicat
ayered User Int
he final user ach layer com
mobility) s and contro rfaces from d
he following the represe ove. The tran
applications ( TL) focuses o ystem, such make transp
L) focuses o ular with res
l for intent‐ n Figure 7) rlying conce mains where ion transpor
erface Model
r interface is mbines three
ls) and data) g elements:
ntation of p nsport layer i (see Section on paramete as timeline ort decision
on the repr pect to the u
aware servic where each pts and mec e user behav rtation as ap
the result o e elements:
ublic transpo s where mo 2.2) operate rs that affec ss and over ns that go b
resentation use of public
ces and app h layer repr chanisms are viour plays a pplication do
of combinati
ortation syst ost currently
e.
ct a user’s e rcrowding. T beyond simp
(data and v
transport sy
plications. resents a
e generic key role. omain for
on of the
tems and available
motional This layer ple route
‐ T o
A featur driven; t this purp of trave layer is transpor persona contexts these lay
3.1
T
3.1.1 M
The tran concerne routes o this data beyond t In GAMB user inte from A t
3.1.2 U
This laye
User Inte
layer com
3.1.3 U
Route fi effective need to approac
The Social Be of people (i.e re of the GA they create a pose, the lay
llers is incre concerned w rtation netw
l trip histor s. In the foll yers in detai
Transport
Motivation
nsport layer ed with key of bus lines a a to compu this layer. BAS we use erfaces that
o B.
UserInterf
er supports o
erface Servic
mputes and
UserInterf
nding is a w e and usable o deviate fr hes for this l
ehaviour Lay e. friends in AMBAS user and exploit yers are stac ementally ex
with informa work provide ies and crow owing, we w l.
Layer
nandObject
serves as a information nd time‐tab ter their se
basic transp allow peopl
faceService
one fundame
ce 1 (Route F
visualise rou
faceSketch
well‐covered visual repre rom establis layer that sh
yer (SBL) focu terms of soc
interface m data to enab cked in a wa
panded from ation about t
er), the high wd data to will describe
tives
foundation about the tr les. Classical rvices, and
port data in t e to make b
es
ental service
Finding): The utes and time
es
use case in esentations. A
shed practic ow how rout
uses on the cial networks model is that ble novel us
y such that m the lowes the transpor her layers e highlight th
the design
of the inten ransport net
route plann most public
two ways: t basic transpo
:
e user specifi etable inform
n transporta As far as the ces. Figure
te informatio
representati s).
t the constit er interfaces the knowled t to the mo rtation netw enrich this b he role of h
principles fo
nt‐aware use work such a ners and trav cly available
o identify tr ort decision,
ies a start lo mation to ge
tion system e Transport L 8 and Figu on can be vis
on of mobili
tuent layers s and provid dge inferred st upper lay work (which basic inform
umans as tr or the user i
er interface s the locatio vel recomme transport a
ansport opti such as wh
cation S and t from A to B
s and most Layer is conce
ure 9 provi sualised on m
ity patterns o
are inheren de novel serv about the b yer. While th is available mation and
ravellers in interface fo
system. Thi on of bus stat
enders heavi applications
ions and to en and whe
d a destinatio B.
applications erned we do ide potentia mobile devic
of groups
ntly data‐ vices. For behaviour he lowest from the integrate different r each of
s layer is tions, the ly rely on don’t go
generate ere to get
on D. The
3.1.4 D
We mak structure all bus li linked to times of about th service o bus stat relevant
3.2
Q
This laye making planners objective
(ad
DataandA
ke use of st e of the tran ines. This en o each other f buses at pa he real‐time or out of ser ions. While t t for predictio
Quality‐of‐
er comprises effective tra s give people
e criteria su
Figure
Figure
dapted from htt
Algorithms
atic and dyn nsport netwo
ables us to l r), possible in articular stat
status of bu rvice) of a b
this layer do on on the hig
‐Transpor
s data, algor ansport deci
e informatio uch as time
8 – Typical visu
(adapted fro
9 – Simple met
tps://play.goog
namic transp ork in terms o learn about nterchange f tions. Dynam uses. These
us, the locat oes not direc gher layers o
rt Layer
rithms and d sions based on about rou of departure
ualization of th om http://live.t
tro map visuali
gle.com/store/
port data o of the locatio
the trajecto facilities as w mic data enr
reports inclu tion of buse ctly involve of our model
design elem
on quality ute options w
e, estimated
he transportatio
transloc.com)
ization on smar
/apps/details?i
n this layer. ons of all sta ries of differ well as to inf iches the sta ude data abo
s, and the e prediction ta l.
ents to prov of transport where route d time of ar
on layer
rtphone
id=us.pandav.N
. Static data ations as wel rent bus line form about d atic data wit out the serv estimated tim
asks, the ava
vide users ri t data. Onlin
recommend rival, travel
NYC)
a describes t l as the time es (which sta departure an th up‐to‐date vice conditio me of arrival ailable data
ich opportu ne and mob dations are
duration, a
the basic etables of ations are nd arrival e reports n (e.g. in l (ETA) at becomes
Howeve often m has unco public tr aspects ‘quality o overed ‘pain ransport. Qu of the user’s of transport’ n objective d nions publish with the cont
e. In dense u nherent exp modes of tr xperience m ourney time lthough stat ng transport QoT measure
NovelConc
er introduces
of‐Transport
nce (such as
of‐Transport
of‐Transport
nalysis of hist
UserInterf
er supports t
erface Servic
mputes and
terface Servi
ed future QoT
UserInterf
mary user in ributes of th ters can be ss of line se sizes. Future
rics neglect ecision base n points’ suc
uality of tra s transport e ’ (QoT) to re data (measu hed on twitt text acquisiti urban areas torical patte
faceService
wo user‐inte
ce 2 (QoT Se
visualise QoT
ce 3 (QoT T
T attributes o
faceSketch
nterface elem he transport
visualised. F egments; Fig
e work will e
important a ed on emotio
h as overcro ansport is a experience th fer to user‐c red by senso er feeds). ion work in W
such as Mad travelling, w n such as ca rolonged wa cted. In the from surveys The objective nd helps use
concepts: at refers to
erface servic
earch): The u T attributes
Temporal Exp
of the transp
es
ment of this network. Fi Figure 10 vi centric meas ors) or on su
Work Packag drid and Lon which is one ars. Overcro aiting times past, data a s has been t e of this laye ers to make t
measures o s).
augmented w o predict Qo
es:
user specifie for routes to
ploration): T port network
layer is the igure 10 and
sualises cro ualises crowd
dvantages a
he travel exp bjective crite delays that m ept that refe eoples’ trans
ures of the t ubjective fee
ge 2 we focu ndon overcro
of the reas owding also caused by c about crowd
taken into a er is therefo travel decisio
of a transpo
with contextu oT attributes
es a start loc nd disadvant
perience and eria. Transpo make people ers to emot
port behavio transport exp
lings (collect
us on crowd owded statio
ons many tr interferes w crowded bus levels has n route segm bus stations tages of each
perience. Qo ted for exam
level as prim ons, buses a
ravellers pre with more tr
s lines migh not been ava ransport ope e user interfa
n QoT param
hat affects t
a.
e point in tim
a destinatio
to explore
ual represen xamples of research ed to use ubjective the term raditional
t lead to ailable to erators in aces that meters.
3.2.4 D
The QoT
‐ f a
‐ f As part sources. The desi this laye Data rep
3.3
P
The Per network
DataandA
T layer relies from direct and/or from ticketin
of this wor
This will en ign of user in
r.
presentation
ersonal
B
sonal Behav k by anticipat
Figure 10
( ad
Figure 1
(adapted f
Algorithms
on quality d measureme
ng informatio k package w nable us to p nterfaces for
and predict
Behaviour
viour Layer ting travel pr
0 – An example
dapted from ht
1 – An example
rom http://hai
ata that can nts of crow
on provided we will inves
predict crow r predicted f
ion algorithm
r
Layer
is designed references (i
e of how a QoT
ttp://content.s
e of how a QoT
irycow.name/c
come from d levels pro
by transport stigate ways wd levels for future scena
ms for this la
to create a ntents) and
T map could be
tamen.com/ze
T map could be
commute_map
two potentia vided by the
t operators. s to integrat a future dat rios is one o
ayer are desc
a personaliz mobility pat
realised.
ro1)
realised.
/map.html)
al sources. e context ac
te and mash ta and time of the key as
cribed in deta
zed experien terns of indi
cquisition fra
h‐up data fr from histor spects of the
ail in Section
nce of the t ividual users
amework
rom both ical data. e work on
n 5.
3.3.1 M
ined subset ble view on ucture, is the r, current tr avioural‐driv ch or irreleva l of this laye tion systems llers. Based rt system can ure mobility rt system, th e for its freq network.
NovelConc
er introduces
l Travel prof
statistical m tion about h time, curren
UserInterf
sonal behavi
terface Servi
ters the use tion for mak
erface Servic
s a time and t information
nandObject
onalized tra y different in oss time and public transp
pular at the
of the tran the transpo erefore not c ansportation ven personal ant informati er is to reve quent usage,
cepts
s three nove
file: this laye model of how ow often an
haviour pred
nd identify e ers in a given ognitive map nsport netwo t location an
faceService
our layer sup
vel informat n daily life. S d space and ortation for weekend. At nsport optio rt system, w consistent w n informatio ization conc on in their c rse the exist s approach is wledge abou d to give peo
establishing ansportation , since passe
l concepts: er records ho w an individua d when an in
diction: a use elements of
situation. p is a visual ork that are m nd transport
es
pports two u
nalized Trav
g access to oices releva
oral Explorat
f interest an travel choice
tion system Specifically, w d b) for diffe
commuting t the same t ons offered which covers with the way on systems a cepts. As a co current trave
ting practice s the discove ndividual use
er transport the route ne
representat most relevan
profile).
user‐interfac
vel Recomme
a cognitive nt to a user
tion of Perso
nd is shown es at that tim
is motivated we can obser erent passen to work, wh travel behav nformation, w motional rela
ecomes muc perceive them
use a public ic transport. es bus stops t profile is u etwork (such
ion of the t nt to a user i
e services:
endation): U e map that
in a proactiv
onalized Trav
a Cognitive
ire transpor and routes o viour, person which is rele
tionship of ch more attr
mselves to b
transport sy A user trans and bus rout used to pred h as bus stop
ransport net n a given con
Using time a summarizes ve manner.
vel Recomm
Map that s ace in the fu
t that people re exist highl nstance, at w
ctivities and in districts o rt network. of the transp the transpor
this principle often burde
ntegrated in the personal nalized view evant to thei passengers ractive and be an integra
ystem over sport profile
tes. dict future t ps, bus lines)
twork that h ntext (deter
and place as s the most
mendations):
summarizes t uture.
es’ travel y diverse weekdays shopping only use a A global portation t system. provides al part of
time and
contains
transport ) that are
highlights mined by
s context relevant
3.3.4 U
The prim possible frequent profile. highlight element predictio week int augment user wit alterativ
3.3.5 D
Key to e of the u knowled algorithm use of c encomp includes
UserInterf
mary user e design ske t routes in t Figure 13 is t elements o ts that are of on it is possi
to the future ted with rea h too much ve design.
DataandA
enhancing th users’ mobili dge about th ms to extract
ontext data asses inform
the source
faceSketch
lement of t etches for a the transpor
on a higher of the transp f less relevan ble to pinpo e. Because o al‐time depar information
F
F
Algorithms
e degree of ty patterns. he user’s typ t the users’ t collected by mation about
and destina
es
his layer is cognitive m rt system an r level of ab port system nce (given th
int transport of the much‐ rture and arr . Future wor
Figure 12 – Des (adapted from
Figure 13 – Des
personalizat Specifically, pical bus usa travel prefer y the data a t the detaile ation station
the cognitiv map. Figure nd is thus a bstraction an
that are of he user’s cur t elements t
‐reduced co rival informa rk will explor
sign Sketch for m http://conten
sign Sketch for
tion of publi , the cogniti age patterns rences from cquisition fr d journeys p ns of their jo
ve map. Figu 12 exploits
direct repr nd shows ho
high relevan rrent tempo that will be o
mplexity of ation, and Q re the advan
Cognitive Map nt.stamen.com
Cognitive Map
ic transport ive travel m . We will th past bus trip ramework on people make ourneys, as
ure 12 and s traces of esentation o ow a transpo nce to a use
ral and spat of importanc a cognitive oT data, wit ntages and d
1 )
2
systems is a ap sketched erefore inte ps. For this p n their mobi e over the co
well as chan
Figure 13 s mobility to of a user’s t ort profile is er while hidi tial context). ce an hour, a map, it can hout overloa isadvantage
a good under d above is b egrate data i urpose, we w ile devices. T ourse of a w nges at inte
how two indicate transport s used to ing those By using a day or a easily be ading the s of each
stops in
identify group of
Social Tr
group is
case differe ential patter ir popularity portance fo hical space may be diffe over time. Th come relevan ere especiall
nating facto sers should b alysed to dis
observed. T nce of the m etails about d
ocial Beh
cial Behaviou s within a use
Motivation
orms are the acknowledge challenge of l settings, pr o have a sign ocial networ and discover rtation has n lish social no blic transpor eased value o
NovelConc
al Behaviour
ravel Profile:
oup uses bu
ravel Behavi
elements of f users in the
ransport Ma
using a publ
nt bus lines n mining alg . A popular r him in hi inside which erent for a g herefore, the nt. This will
y working d r. Temporal be presented
cover popul he predictio mobile travel data and algo
aviour La
ur Layer is er’s social ne
nandObject
e standards w ed that makin f promoting roviding peo nificant effec rks such as F ring content not been exp
iour Predicti
f the route e future.
ap: a social t lic transport
are taken as gorithms, rev route is one
designed to etwork.
tives
we use to jud ng pro‐envir g sustainable ople with evi
ct on behavio Facebook an t among frien plored yet. T
ic transporta encourage tr
information
duces three avel profile e bus routes.
ion: a social network (su
ransport ma system.
s part of a tri vealing inform e that the us
The popula
mobility ta ted informat which a den will be used
the personal
o create aw
dge the appr ronmental so
e behaviour idence of w our, for exam
d Google+ a nds. Howeve The social be ation and ma ravellers to r n system.
novel conce encompasse
travel profi uch as bus s
ap is a visua
ip. The trip h mation abou ser frequentl ar routes w kes place. H s constitutes e an associat ser interface sure activitie habits can al tion. Therefo nse concentr
then to info
behaviour la
areness of
ropriateness ocial norms m ehaviour laye ake people a rely on publi
epts.
es informatio
le is used to tops, bus lin
l representa
history will b ut the routes y travels on will be explo However, sin s dynamic k tion to the d
to different s at the wee
travel prefe
of our own more visible tory studies around them of public tra built upon th
cation of thi er encompas
ware of how knowledge w
days and tim t structures i
ekend are a as triggers t r trip historie milar departu ntrol logic a
found in Sec
erences and
actions, and e is an impor s and more m are doing
ansport. he social met
ehaviour (int e most relev
ghlights how
asic input ndividuals fore, is of
mobility
it is now rtant part applied, has been
taphor of to public elements
3.4.3 U
The soci
User Int
paramet network
User Inte
interest predicte
3.4.4 U
The prim 14, Figur overlap number transpor aggregat will expl
UserInterf
al behaviour
terface Serv
ters the user k is using a pu
erface Servic
and is show ed to be using
UserInterf
mary user int re 15 and Fig in the areas
of users (in rt traces of tion of cogni ore the adva
faceService
r layer suppo
vice 6 (Socia
r is getting ac ublic transpo
ce 7 (Tempor
wn a social g a public tra
faceSketch
terface elem gure 16 show
of interest this case jus
members o itive maps o antages and
Figure
Figure (ada
es
orts two user
al Transport
ccess to a so ort system.
ral Exploratio
transport m ansport syste
es
ment of the s w possible de
(defined by st two). Figu of the user’s
f individual u disadvantag
14 – Envisioned
15 – Envisione apted from sou
r‐interface se
t Usage Dis
ocial transpor
on of Social T
map that su em at the ind
social behav esigns of a so
the reachab re 15 shows s social netw users (again, ges of each a
d design of a So
ed design of a S rce: http://ma
ervices:
scovery): Us rt map that s
Transport Us
mmarizes h dicated time
iour layer is ocial transpo ble travel des
a social ma work. Figure , in this exam lterative des
ocial Transport
Social Transport cwright.org/ru
ing time an summarizes
sage): The us ow the use .
the social tr ort map. Figu stination wit p that is gen e 16 shows mple just two sign.
t Map 1.
t Map 2 unning/)
nd place as how the use
ser specifies r’s social ne
ransport ma ure 14 repre thin short tim nerated by o
a social ma o users). Fut
s context er’s social
a time of etwork is
3.4.5 D
The soci Therefor informat persona social la available this laye Howeve policy ru data can users. In order required habits b the proc the user algorithm routes h destinat transpor the user More de
DataandA
ial layer targ re, it has the tion. From a l layer descr yer deals wi e on the dev er includes re
r, the sharin ules, where u n be retrieve
to explore t d. In order to etween the cessing and c r’s identity a ms on the pe have to be d ions in rela rtation netw r’s interface. etails about d
Figure
Algorithms
gets the tra e most comp a data proce ribed previou
ith a set of d ice of the qu emote acces ng of travel users state t ed which bel
the travel ha o address th
users. This communicat and the rou ersonal laye displayed. Fo ation to the work with clu
data and algo
16 – Envisione
vel behavio prehensive d essing point usly. While t different, bu uerying user, ss mechanism
data obeys heir prefere ongs to user
abits of socia is issue, we is done to a ion load on tes most oft r. The social or instance, e trips of a ues on the fr
orithms for t
ed design of a S
ur of group ata requirem of view, the the personal ut possible r , it has to be ms to coord
precise priv nces of who rs who have
al friends, in exchange o avoid transfe the mobile d ften taken b
l layer will th this could a user. The riends’ trave
the personal
Social Transport
s who have ments conce e social layer
layer extrac related trave e retrieved fr inate the on vacy restrict o else may ac defined acc
nformation a nly a summa erring low‐le
devices. The by this user, hen use app
be those ro e routes can el behaviour
behaviour la
t Map 3
strong soci rning the am r represents cts the users el patterns. S
om the user n‐demand da tions as defi ccess their d cess permissi
about their r ary of the m vel trip histo e retrieved tr as inferred ropriate met outes having
n then be and will be
ayer can be f
al ties amon mount of use
s an extensio s’ travel patt
Since this da r’s friends. Th
ata retrieval ined by use ata. Hence, ions for the
regular route most significa ories so as t ravel data co
from the p trics to deci g similar sou used to en visualized a
found in Sec
ng them. er‐related on to the terns, the ata is not herefore,
process. r‐specific only that querying
es will be ant travel to reduce onsists of prediction de which urces and nrich the as part of
4
Sy
Over the concepts views on required integrati
4.1
U
The arch shown in processi mobile u in our sy architect
4.1.1 P
The pres various pattern. to requ informat (cognitiv behaviou that the change p
stem
Arc
e course of s of the use n this system d to implem
ion of the int
User
Interf
hitecture of n Figure 17.
ng and disp user device p
ystem is sho ture more in
Presentatio
sentation lay travel infor
The presen est and ret tion access a ve map, soci
ur‐driven inf view looks preferences
chitectur
the project, r interface m m. First, we d
ent our syst tent‐aware i
face
Syste
our user in It is based o laying of co participating own in Figur n detail.
Figur
on
yer impleme mation. Hen tation comp trieve desir and cognitive ial map, etc. formation ab
different for settings, suc
re
, we will de model prese discuss an int
tem. Then, w nterface syst
em
nterface syst n a set of dif ntext data.
in the GAM re 17. In the
re 17 – Archite
nts the inter nce, it corre ponent provi ed travel in e processing .). The interf bout the tra r every user ch as the priv
evelop a mo ented before ternal persp we present tem into the
tem follows fferent funct Instances of MBAS system
e following,
cture of the Us
rface to the u esponds to t ides interact nformation. g as easy as p
face will not avel habits o in the syste vacy policies
obile travel s e. In this sec ective which
an external e overall arch
s a model‐vi tional compo f the archite
. The data fl we present
ser Interface Sy
users and is the view pa tive user inte The interfa possible, usin t only includ of each user. m. The comp
stating the d
system whic tion, we intr h represents perspective hitecture.
iew‐controlle onents, whic ecture will be
ow among d
the compo
ystem
focused on t art in the m erface eleme ace will be ng the conce de static tran
. Therefore, ponents will data sharing
ch integrates roduce two the key com e and talk a
er design pa ch manage a
e deployed different com onents of ou
the represen model‐view‐c ents that all e designed
epts describe nsport data,
we like to p
also enable g restrictions
s the key different mponents bout the
attern as accessing, on every mponents ur system
ntation of controller ow users to make ed earlier
but also point out e users to
4.1.2 C
The con this inte changes events f represen user. On user to t needs o instance at this st
4.1.3 D
The data the data user’s c changes statistica and muc derived deciding
Controller
troller medi eraction, the in the user from the p nts a reactive n the other h trigger autom of the user t e, by noticing tation can be
DataInfere
a inference c a available i
ontext to g will be proc al representa ch more co
which are th g when to sho
Figur
ates betwee e controller
r interface a presentation e strategy fo hand, the co matic change that may re g that the us e delivered t
ence
component n the GAMB ather inform cessed to fit ation. The st mpact than hen used to ow travel‐re
re 18 – User Int
en the prese is intended are reactive component or adapting t ontroller also es in the use equire specif er is heading o the user.
consists of a BAS system. mation abou t a stochastic
tochastic mo the origina personalize lated inform
terface System
entation and to act in a or proactive t to retriev the user inte o continuous er interface. I fic informati g towards to
a set of algo
The compo ut the user’
c model whi odel is an ab
l user trace the user int mation.
m Data Flow Dia
d the data p
two‐way d e in nature. ve and man erface since t sly observes In this case, ion and a p o a bus statio
rithms for m onent contin
s daily rout ich summari bstract repre
s. Based on erface view
agram
roviding com irection, dep On the one nipulate the
the interacti relevant co the controlle roactive stra on, the arriva
mining behav nuously mon ine. The tra zes the past esentation o
the model, and create p
mponents. R pending on e hand, it in e current vi
ion is initiate ntext chang er anticipate ategy is cho al times of ne
vioural patte nitors change
aces of past t user behav of the user b , travel patt proactive tri
Regarding whether ntercepts ew. This ed by the es of the es further osen. For ext buses
4.1.4 D enables decide w private. he can s users to the data user inte interface privacy defined map wil agreed o
4.2
In
The inte GAMBAS
DataAcces
r interface sy bile travel ap
g local and r mobile device
tion in form t stakeholde k provider, e e friends in a tive, this com 1.
Experience
erience logg travel applic s and consu
und of the a the logs as p e, based on a
when search cation and tion. Therefo ystem and to
Privacy
ng the users hrough ena ng the priva e access righ e‐processable
users to disp what contex
For instance simply use th
create perm a as well as th
erface intera e, which will S middlewar
s
ystem is hea pplication. Fo remote data e, comprisin
of past user ers in the GA xternal aggr a social netw mponent bui
eLogging
ing compone cation. Basic mes differen application a part of our e xtual‐informa e, in case a u
he interface mission relati
he social gro acts with the
be defined n specificatio verable will b de informati he data withi
n into GAM
nterface syst e. These inte
avily data‐dri or this purpo sources. Loc g real‐time c r traces. In c AMBAS syste regation serv work. Since t
ilds on the p
ent comes w cally, it logs nt types of and intercep evaluation pr of the intera ded informa nterface con sents an imp ssible improv
a major goal having cont has been h
MBAS arc
tem require eractions wil
iven and use ose, a compo cal data refe context data contrast, the em. They inc portant instr vements.
of the GAM trol over th as been sta ta. However, ectly for end heir individu e revealed to
to not share the sharing f nvolve the ty which is gra eservation fra
he external d teworthy to cted by the he transport al network.
chitecture
es frequent ll take place
es knowledg onent is req ers to the con
, e.g., the cu e remote dat lude service r crowd‐leve o remote dat security mec
tools to mon t how the n. The comp nt interaction
sess the valu we can unde e same time
ost effective rument to in
MBAS middlew heir persona
ted in D3.1.
function. The ype of contex
nted access amework th dependencie o mention th
policy config t usage of th
e
interactions upon well‐d
e from vario uired to pro ntext inform urrent locatio
ta sources ar s and databa el statistics a ta is highly se
hanisms tha
nitor pattern user interac onent is des ns observed ue of our int erstand how
, we can ve e to provid nform an iter
ware. We re al informatio
1, where po policy langu refore, the u olicies. This m
d what info rt informatio e user interf guration. For hose friends
with other efined interf
ous sources t ovide mechan mation direct
on, as well a re distribute ases of the t as well as tra ensitive from t have been
ns in the usa cts with the signed to ru at run‐time terface conc w much time rify in which de users wit rative design
esolve privac on. The con olicies are a uage is desig user interfac means that u
rmation will on on weeke face therefo red, the gran
component faces provide
transport avel data mponent e second concepts the social explicitly
middlew depicted
As the fi support distribut GAMBAS abstract in a unifo The data available about th residing framewo to travel and real ridership bus rout user inte The inte supporte transpor
ware compon d in Figure 19
F
igure shows, interactions ted in the GA S query pro ion, so that t orm manner a acquisition e on the loc he real‐time
at remote ork in GAMB llers. The tra
‐time bus st p histories an tes of a user’ erface has to ent‐aware u ed by the qu rt schedules.
nents. The i 9.
igure 19 – Inte
, each mobil s with a mob AMBAS syste ocessing sys the access is r.
n framework cal semantic context of
parties is BAS. Various ansport netw atus update nd provides ’s friends, is o gather this ser interface uery process . In contrast,
ntegration o
gration of User
e device is e ile traveller. em. The user stem. The q
provided tra
k provides pe data storag users (e.g. m accessed th GAMBAS se work provide s. Further, a access to fut hosted on th information e system us sing system. , continuous
of our syste
r‐Interface Syst
equipped wi This require r interface sy query proce
ansparently
ersonal data ge and can b
mode of cur hrough the ervices are in er offers serv a GAMBAS ag
ture crowd l he personal
from the GA ses both the
One‐time q
queries are
em into the
tem into GAMB
th the inten es access to v
ystem will co essing system
from their a
a about a us be accessed rrent travel)
privacy pre nvolved in m vices to infor ggregation s
evel predict devices of ea AMBAS syste
e one‐time queries are u
used to dea
overall GAM
BAS Architectu
t‐aware use various sourc ontact these m provides ctual locatio
ser’s activitie directly. It as well as eservation a
anaging dat rm about the
ervice will be ions. Social t ach of the fr em in a distri and continu used to requ al with dynam
MBAS archit
re
r interface s ces of data w e sources thr
a high lev on in the netw
es. This data includes info past user tr and data pr ta that are of e transport s
e deployed t travel behav riends. There ibuted mann uous queryin uest static d mic data give
ecture is
system to which are rough the vel query work and
a is made ormation rips. Data rocessing f interest schedules to collect viour, i.e., efore, the ner.