Conference PaperPDF Available

A Model and Framework for Matching Complementary Spatio-Temporal Needs

Authors:

Abstract and Figures

Currently, systems that let people search for opportunities to fulfill their spatio-temporal needs are built according to the conceptual model of service provider and consumer: After the providers make their needs publicly available, consumers use a specifically tailored query engine to find fitting offers. E.g., in carpooling, someone wants to fill an empty seat and to share costs (and publishes this offer), while another person wants to travel the same route. This model prevents the consuming side from making their needs available to the service providers and makes it hard to generalize, as query engines require rigid (often domain-specific) properties. Addressing this problem, we propose a generic model for publishing and processing complementary spatio-temporal needs. Our model uses a simulator to assess how well the collaboration between different entities would approximate their goals. To reuse existing concepts and embed the model into the emerging Semantic Web, everything is modeled in accordance with Linked Data principles.
Content may be subject to copyright.
A Model and Framework for Matching
Complementary Spatio-Temporal Needs
Dominik Bucher
ETH Zurich, Institute of
Cartography and Geoinformation
Stefano-Franscini-Platz 5
Zurich, Switzerland 8093
dobucher@ethz.ch
Simon Scheider
Utrecht University, Department of
Human Geography and Planning
Heidelberglaan 2
Utrecht, Netherlands 3584 CS
s.scheider@uu.nl
Martin Raubal
ETH Zurich, Institute of
Cartography and Geoinformation
Stefano-Franscini-Platz 5
Zurich, Switzerland 8093
mraubal@ethz.ch
ABSTRACT
Currently, systems that let people search for opportunities to fulll
their spatio-temporal needs are built according to the conceptual
model of service provider and consumer: After the providers make
their needs publicly available, consumers use a specically tailored
query engine to nd tting oers. E.g., in carpooling, someone
wants to ll an empty seat and to share costs (and publishes this
oer), while another person wants to travel the same route. This
model prevents the consuming side from making their needs avail-
able to the service providers and makes it hard to generalize, as
query engines require rigid (often domain-specic) properties. Ad-
dressing this problem, we propose a generic model for publishing
and processing complementary spatio-temporal needs. Our model
uses a simulator to assess how well the collaboration between dif-
ferent entities would approximate their goals. To reuse existing
concepts and embed the model into the emerging Semantic Web,
everything is modeled in accordance with Linked Data principles.
CCS CONCEPTS
Information systems Location based services;Content
ranking;Web and social media search;
KEYWORDS
ACM proceedings, spatio-temporal, modeling, matching, linked data,
ontology
ACM Reference format:
Dominik Bucher, Simon Scheider, and Martin Raubal. 2017. A Model and
Framework for Matching Complementary Spatio-Temporal Needs. In Pro-
ceedings of SIGSPATIAL’17, Los Angeles Area, CA, USA, November 7–10, 2017,
4 pages.
https://doi.org/10.1145/3139958.3140038
1 INTRODUCTION
Imagine you are searching for an aordable apartment, a carpool-
ing partner, or a jazz concert next weekend. Commonly, you would
use one of the many existing websites to nd opportunities that
satisfy your needs. Most of these platforms are built according to
Permission to make digital or hard copies of part or all of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for prot or commercial advantage and that copies bear this notice and the full citation
on the rst page. Copyrights for third-party components of this work must be honored.
For all other uses, contact the owner/author(s).
SIGSPATIAL’17, November 7–10, 2017, Los Angeles Area, CA, USA
©2017 Copyright held by the owner/author(s).
ACM ISBN 978-1-4503-5490-5/17/11.
https://doi.org/10.1145/3139958.3140038
the conceptual model of service provider and consumer: One party
oers something, while another has a complementary demand, so
that working together results in a benet for both. It is commonly
required that the oering side rst makes its oer publicly avail-
able, and only a simple (yet domain-specic) query engine inter-
face is provided to the demanding side.
These current systems exhibit the following limitations:
They are either highly specialized and thus rigid in their
domain of application (e.g., job platforms), or too generic
to match specic needs (e.g., portals for classied ads), in
which case users are barely supported during their search.
Building highly specialized platforms often results in a “keep-
ing of data in silos”, which makes interaction with such sys-
tems (for users and automated services) dicult (cf. [12]).
There is no mechanism to account for the earliest possible
matching of needs. The requirement that providers have to
explicitly publish oers on the platform, before demanding
parties are able to search for them, introduces a considerable
delay between the posting of a need and a possible match.
As a result, users have to learn how to interact with a wide range
of platforms, there is still no generic search engine that answers a
“search for a yellow car for sale in Massachusetts” [1, p. 179] in a
meaningful way, and it remains dicult for the consuming side
to publish their needs (e.g., making the statement “I am looking
for a ride to Rome” available to potential carpooling providers). A
system overcoming these limitations would be highly desirable, as
matching the needs of people is a common task bounded by spatio-
temporal constraints and limited availability of results (cf. [16]).
In this paper, we introduce a model and framework that species
the components, concepts and algorithms needed in order to pro-
cess spatio-temporal match queries in real-time, making optimal
and early matches possible (cf. [3]). Such a model may constitute
a fundamental component for a variety of systems, e.g., coopera-
tive transportation systems [13] or tools that encourage people to
engage in a more sustainable mobility behavior [21].
2 RELATED WORK
Studies showed that around 20% of all Web search engine queries
have a geographical intent [18]. The implicitly stated needs can be
for habitation, accommodation, goods, cars, jobs, etc. [9], and often
involve commercial aspects (“buy or rent”) which constitute im-
portant aspects of future Geographical Information Retrieval (GIR)
search engines (cf. [14]). Looking at systems and applications that
focus on improving the state of the art in spatio-temporal need
SIGSPATIAL’17, November 7–10, 2017, Los Angeles Area, CA, USA D. Bucher et al.
matching, we mostly nd specialized services and crowdsourcing
applications (cf. [4]), tackling dicult domains such as individual
transport suggestions. Examples include recommendation of sus-
tainable transport solutions [6], or real-time matching for carpool-
ing [11]. Reviews of such specialized services reveal that users are
interested in a variety of attributes of the involved objects (e.g.,
number of rooms in a house), and that systems should allow for
spatial and temporal exibility [5]. It is also noted that an open
and extensible technological infrastructure is missing [7].
Raubal and Winter [15] propose a generic way towards ad-hoc
collaborative decision making which uses agent planning to de-
velop strategies that yield desirable outcomes. Planning (as a sub-
eld of articial intelligence) is concerned with nding action se-
quences to reach a certain goal state [19]. In our case, we are pri-
marily simulating how a graph (with global knowledge) evolves
over time, based on dierent actions the agents can perform. Such
evolving graphs are common in many domains, including trans-
port and mobility [8], and several algorithms have been adapted
to search and process them [17]. We model these graphs as Linked
Data, a term that denotes data on the World Wide Web which is not
self-contained, but semantically linked to other information [2].
We argue that basing the model on Linked Data is benecial, as
this makes integration into the Web easy, and allows systems to
access and process needs automatically.
3 MODELING COMPLEMENTARY NEEDS
Every need is an expression of a certain goal: Some agent wants
a certain state to become true, but cannot be sure about its feasi-
bility, which is determined by other entities capable of cooperat-
ing, or resources to trade and consume. While some of these fac-
tors are changing in space and time (e.g., a car drives around, and
thus might be available at some time, but unavailable at another),
many are spatially invariant (a shop does not change its location
once built, but its stock changes frequently), or even completely
static (a certain car type always has the same number of seats). In
our model, we denote all this (formalized) knowledge as the world
knowledge base (world KB).
3.1 Modeling States with Linked Data
We build our model in alignment with Linked Data [2] principles,
due to them being conceptually simple, generally applicable, and
linked, which results in a large distributed net of knowledge that
we can reuse. The fundamental concept of our model is a state
s=(t,P,N), which is a momentarily snapshot of some statements
within a graph, where tRis a time associated with this state,
Pis a set of positive statements, and Nis a set of negative state-
ments (as Linked Data makes the open world assumption, we need
to encode these statements separately). Both positive and negative
statements are triples, built from resource identiers, literals, or
blank nodes.
States can be merged into an existing Linked Data graph, to cre-
ate a view, i.e., a snapshot graph which describes statements that
hold true at t. To compute the view, we take an arbitrary graph,
and merge a number of states siS, sorted by their times ti, into
it. When merging a state siinto a graph, rst, all negative triples
Ni(accepting any node in position of an empty placeholder) are
removed, after which the positive triples Piare added to it. This
allows to remove or “overwrite” certain triples, for example the
location of a subject (otherwise, there would be two triples speci-
fying the location).
Figure 1: An example of two arbitrary states, specied as
named graphs in Linked Data. The second state s2overwrites
the location property of car, which is done by specifying the
latitude and longitude as negative statement triples.
We use the blank node _:self to describe meta information about
the state itself, such as the time when it occurs, or its type. Figure 1
gives an example of two states, where the second state overwrites
statements of the rst at a later point in time.
3.2 Processing Spatio-Temporal Needs
States appear within the world KB (describing changes to general
knowledge at certain points in time) and within spatio-temporal
needs n=(F ,G), where they describe the discrepancy between
the way things are (facts F) and the goals Gsome entity has. While
fact states fi∈ F contain everything that is important during the
matching process (in addition to the general knowledge within the
world KB), the goal states are what should be approximated when
trying to resolve the need. Ultimately, we want to approximate all
the views created by adding the goals of a certain need to the world
KB: For each goal дi∈ G, we add the states of nto the world KB
(SwK B ∪ F ∪ {дi}), and evaluate the corresponding world view
at time ti(of дi), resulting in a “hypothetical” graph in which this
particular goal holds true.
To evaluate these hypothetical graphs against possible outcomes
of agents collaborating, we introduce actions, which are known
constructs from planning literature. In our case, an action is dened
as a=(R,p,e), where ri(w,n1,n2, ... ) ∈ Rare retrieval functions,
returning sets of actionable elements Ae, based on the world KB w
and a set of needs niN. The precondition function p(Ae), which
evaluates to true or false, determines whether the action is appli-
cable to a set of Ae, and e(Ae,w,n1,n2, ... )computes the eects of
applying the action, which produces a set of result states Sa.
Listing 1 shows a retrieval function which species the retrieval
of a a rider (an actionable element), used within a carpooling action,
as a SPARQL query. It retrieves a possible passenger by searching
for needs that specify someone has the goal to reach a certain lo-
cation (in state s2), starting from a known location (in state s1).
SELECT DISTINCT ?rider ?rider_time_start ?rider_time_end
?rider_lon_start ?rider_lat_start
?rider_lon_end ?rider_lat_end
WHERE {
?need a n:Need .
?need n:state ?s1 .
?need n:state ?s2 .
Matching Spatio-Temporal Needs SIGSPATIAL’17, November 7–10, 2017, Los Angeles Area, CA, USA
GRAPH ?s1 {
:self a n:FactState .
:self n:dateTime ?rider_time_start .
?rider wgs84_pos:long ?rider_lon_start .
?rider wgs84_pos:lat ?rider_lat_start .
}
GRAPH ?s2 {
:self a n:GoalState .
:self n:dateTime ?rider_time_end .
?rider wgs84_pos:long ?rider_lon_end .
?rider wgs84_pos:lat ?rider_lat_end .
}
FILTER (?rider_time_start < ?rider_time_end)
}
Listing 1: Querying for a rider, as part of a carpooling action.
In the case of carpooling, p(Ae)ensures that the start and end
locations of the car and the rider match (within some buer area),
and could possibly be extended to ensure that the number of pas-
sengers does not exceed the number of seats, or that a rider has a
minimal amount of money available. Listing 2 shows the relevant
part of e(·), where we simply replace all variables with the respec-
tive elements of Ae. This generates a state describing the outcome
if ?rider chooses to travel in ?car.
GRAPH :g {
:self a n:FactState ;
n:dateTime ?car_time_end .
?car :passenger ?rider .
?rider wgs84_pos:long ?car_long_end ;
wgs84_pos:lat ?car_lat_end .
}
Listing 2: Eects of applying a carpooling action.
3.3 Spatio-temporal Optimizations
Evaluate each need in combination with every other need and ac-
tion would essentially require |N|2· |A|computations for single-
action plans involving two needs only. As all the needs are of a
spatio-temporal nature, we can use the following properties for
optimization:
Candidate needs have to involve some objects that are spa-
tially close. The denition of closeness depends on the ac-
tion and the need itself (e.g., nding an apartment is usually
possible in a much larger area than nding a restaurant).
The states of involved needs have to be temporally close.
Again, this denition depends on the action and the need
itself.
Only certain needs are valid candidates for an action. As
such, it is not required to evaluate impossible combina-
tions of actions and needs.
Ultimately, we denote a set of needs to be complementary, if
there exists an action awith Ae,,p(Ae)=true , and where there
exists at least one spatial and temporal overlap in the states of any
ni,njN,i,j. Note that our denition of complementarity is
motivated by selecting a set of candidate needs which potentially
can be used to fulll a persons need.
3.4 Simulating how Entities Work Together
Corresponding to the three points above, three indexes store ref-
erences to needs: (1) a spatial index Is, indexing on all subjects ap-
pearing within a need which have a location given, (2) a temporal
index It, indexing on the temporal extent of dierent states, and
(3) a lookup table In, which stores needs with Ae,for all ac-
tions. To evaluate how well needs complement each other, we use
the following procedure upon a new need nnentering the system:
(1) Indexing. The retrieval functions rjRof all actions aiA
are applied to nn, resulting in sets of actionable elements
Ae,i. If Ae,i,, need nnis added to In[ai](i.e., nnbecomes
a candidate for action ai). All times tof the states Fand Gof
nnare added to It(with tolerances). All states (in F ∪G ) are
searched for triples describing locations of subjects which
are then added to Is.
(2) Complementary Candidate Retrieval. Using the indexes Is,It
and In, a set of candidate needs Nc=In[ai] ∩ Is[tTn] ∩
It[lLn]is retrieved (Is[tTn]and It[lLn]yield entries
overlapping with times and locations of nn).
(3) Simulation of Outcome. For each need ncNcin combina-
tion with nn, the cross product of all Aeis built, which yields
all combinations of actionable elements with respect to ac-
tion ai. For each element produced by the cross product, we
evaluate the eect ei, in case the precondition pievaluates
to true, and add the resulting state set to the set of possible
outcomes O.
(4) Evaluation of Outcome. Finally, the similarity between the
views given by these outcome states and the views produced
by the original goal states Gis computed, and the resulting
ranked list is returned to the user.
While arguably this last step is action-dependent, we can still
state that the more similar a result state to a goal state is, the better
the action satises the needs. In our implementation, we opted for
a simply structural and lexical similarity measure, as introduced
by Zhang et al. [22]. Note that optimally, the similarity assessment
would be depending on users’ preferences, either specied manu-
ally, or learned automatically [10].
4 CASE STUDY
We implemented the above introduced framework in Python, us-
ing the rdf library1for all Linked Data and SPARQL parts, and per-
formed a case study on carpooling needs (heuristically generated
using properties of crawled data from carpooling platforms). We
selected one carpooling oer as the “target” of the query presented
below, and copied it four times, modifying several properties of the
copies, to make them deviate slightly. When querying the system
with a need perfectly matching the target, we expect the highest
outcome score for the target itself, with the slightly modied needs
following up behind.
Current carpooling websites allow specifying the locations from
and to, and the date of travel. The resulting list contains the time
of travel, the diver’s name and rating, the exact locations, the price,
and some other metadata which has to be manually assessed. To
add a dimension, let us assume the traveler brings a large bag, and
thus needs a spacey car. Using our framework, this need could
be formalized with the following goal (the fact state denoting the
starting location is omitted):
:AlicesGoal {
:self a n:GoalState ;
1https://github.com/RDFLib/rdib
SIGSPATIAL’17, November 7–10, 2017, Los Angeles Area, CA, USA D. Bucher et al.
Table 1: Top 5 results of the query presented in the transport
case study.
Similarity Arrival Location Time Luggage Room
0.8187 (8.30, 47.05) 16:42 800 liters
0.8163 (8.50, 47.25) 16:42 800 liters
0.7451 (8.30, 47.05) 16:42 300 liters
0.7427 (8.50, 47.25) 16:42 300 liters
0.7299 (8.50, 47.25) 19:00 300 liters
n:dateTime "2017-07-07T16:42:00Z"^^
xml:dateTime .
:Alice wgs84_pos:long 8.304375 ;
wgs84_pos:lat 47.049545 .
_:anyCar a :Car ;
n:luggageSpaceLiters 800 ;
n:hasPassenger :Alice .
}
Listing 3: Alice’s ultimate goal when looking for carpooling.
Table 1 shows resulting matches. The similarity column denotes
how similar the outcome graph (after applying the carpool action)
is to the above goal. It can be seen that the two top results both
are from cars with a luggage space of 800 liters, where the second
one arrives at a dierent location than specied in Alice’s goal.
The three following needs are all oering a smaller luggage com-
partment, and the only one with a dierent arrival time is ranked
lowest.
5 CONCLUSION AND OUTLOOK
In this paper, we proposed a generic model for location-based search
that enables people to publish and match their complementary
spatio-temporal needs. We formalized the notion of complemen-
tary need and introduced an agent planning procedure to nd po-
tential matches. There are a number of open questions that need
to be addressed in the future, such as the challenges arising from
the usage of Linked Data for representing and publishing needs.
Real-time matching: Due to the high volatiliy of data, in-
formation providers should be proactively updated by push-
ing new needs to interested matching systems.
Removing fullled needs: Once agents have decided to
collaborate, the respective needs need to be removed from
the system.
An ontology to specify needs: It is important to agree on
some terms to make information accessible and machine-
processible (cf. schema.org).
Further research questions arise from modeling goal states and
actions. In our model, the eect functions of actions can have ar-
bitrary complexity, which also means they have to be specied
for a lot of cases individually. It would be interesting to nd out
if and how process ontologies can be used to provide design pat-
terns, e.g., for needs with an oer / demand structure. Finally, both
the degree of complementarity (which is dependent on user pref-
erences, which are not captured within our framework yet) and
privacy-related topics (cf. [20]) will further have to be considered
and integrated into the proposed framework.
Acknowledgements. This research was supported by the Swiss
National Science Foundation (SNF) under the National Research
Program NRP71 “Managing Energy Consumptions”. Additional sup-
port is acknowledged from the Swiss Competence Center for En-
ergy Research (SCCER) Ecient Technologies and Systems for Mo-
bility and the Commission for Technology and Innovation (CTI).
REFERENCES
[1] Tim Berners-Lee and Mark Fischetti. 1999. Weavingthe Web: The Original Design
and Ultimate Destiny of the World Wide Web by Its Inventor (1st ed.). Harper San
Francisco.
[2] Christian Bizer, Tom Heath, and Tim Berners-Lee. 2009. Linked data-the story
so far. Semantic services, interoperability and web applications: emerging concepts
(2009), 205–227.
[3] Dominik Bucher, Paul Weiser, Simon Scheider, and Martin Raubal. 2015. Match-
ing Complementary Spatio-Temporal Needs of People. In Web Proceedings of the
12th International Symposium on Location-Based Services.
[4] Lei Chen and Cyrus Shahabi. 2016. Spatial Crowdsourcing: Challenges and Op-
portunities. IEEE Data Eng. Bull. 39, 4 (2016), 14–25.
[5] Zhirong Chen, Jianhong Xia, and Craig Cauleld. 2014. A survey of a person-
alised location-based service architecture for property hunting. Journal of Spa-
tial Science 59, 1 (2014), 63–78.
[6] Sergio Di Martino, Raaele Galiero, Clemente Giorio, Filomena Ferrucci, and
Federica Sarro. 2011. A Matching-Algorithm based on the Cloud and Positioning
Systems to Improve Carpooling. In DMS. 90–95.
[7] Daniel Graziotin. 2013. An Analysis of issues against the adoption of Dynamic
Carpooling. CoRR abs/1306.0361 (2013).
[8] Peter Grindrod and Desmond J Higham. 2010. Evolving graphs: dynamical mod-
els, inverse problems and propagation. In Proceedings of the Royal Society of Lon-
don A: Mathematical, Physical and Engineering Sciences, Vol. 466. The Royal So-
ciety, 753–770.
[9] Andreas Henrich and Volker Luedecke. 2007. Characteristics of Geographic In-
formation Needs. In Proceedings of the 4th ACM Workshop on Geographical Infor-
mation Retrieval (GIR ’07). ACM, New York, NY, USA, 1–6.
[10] Krzysztof Janowicz, Benjamin Adams, and Martin Raubal. 2010. Semantic Refer-
encing - Determining Context Weights for Similarity Measurement. Geographic
Information Science 6292 (2010), 70–84.
[11] Chih-Hsiang Lin, Ming-Kai Jiau, and Shih-Chia Huang. 2012. A cloud comput-
ing framework for real-time carpooling services. In Information Science and Ser-
vice Science and Data Mining (ISSDM), 2012 6th International Conference on New
Trends in. IEEE, 266–271.
[12] Grant Donald McKenzie. 2015. A Temporal Approach to Dening Place Types
based on User-Contributed Geosocial Content. Ph.D. Dissertation.
[13] Harvey J. Miller. 2013. Beyond sharing: cultivating cooperative transportation
systems through geographic information science. Journal of Transport Geogra-
phy 31 (July 2013), 296–308.
[14] Ross Purves and Christopher Jones. 2011. Geographic Information Retrieval.
SIGSPATIAL Special 3, 2 (July 2011), 2–4.
[15] Martin Raubal and Stephan Winter. 2010. A spatio-temporal model towards ad-
hoc collaborative decision-making. In M. Painho, M. Santos, and H. Pundt (Ed.),
Geospatial Thinking. Lecture Notes in Geoinformation and Cartography. Springer,
Berlin, Heidelberg, 279–297.
[16] Martin Raubal, Stephan Winter, Sven Tessmann, and Christian Gaisbauer. 2007.
Time geography for ad-hoc shared-ride trip planning in mobile geosensor net-
works. ISPRS Journal of Photogrammetry and Remote Sensing 62, 5 (2007), 366–
381.
[17] Chenghui Ren, Eric Lo, Ben Kao, Xinjie Zhu, and Reynold Cheng. 2011. On
querying historical evolving graph sequences. Proceedings of the VLDB Endow-
ment 4, 11 (2011), 726–737.
[18] Mark Sanderson and Janet Kohler. 2004. Analyzing geographic queries. In SIGIR
Workshop on Geographic Information Retrieval, Vol. 2.
[19] Ioannis Vlahavas and Dimitris Vrakas. 2005. Intelligent techniques for planning.
IGI Global.
[20] Paul Weiser and Simon Scheider. 2014. A civilized cyberspace for geoprivacy.
In Proceedings of the 1st ACM SIGSPATIAL International Workshop on Privacy in
Geographic Information Collection and Analysis. ACM, 5.
[21] Paul Weiser, Simon Scheider, Dominik Bucher, Peter Kiefer, and Martin Raubal.
2016. Towards sustainable mobility behavior: Research challenges for location-
aware information and communication technology. GeoInformatica 20, 2 (2016),
213–239.
[22] Dehai Zhang, Tianlong Song, Jun He, Xingwei Shi, and Yan Dong. 2012. A
similarity-oriented RDF graph matching algorithm for ranking linked data. In
Computer and Information Technology (CIT), 2012 IEEE 12th International Confer-
ence on. IEEE, 427–434.
... An example of code to define an event is shown in Figure 5. Alternatively, more extensive ontologies can be found in, for example, Bucher et al. (2017) or Scheider and Tomko (2016). Not all spatial operators identified in the previous section are essential for our spatial CEP. ...
Article
Full-text available
The types of data available have changed in the last decade. While, historically, data were gathered in batches and distributed as such, e.g. as a database or shapefile, today we are dealing increasingly with real-time data. This data is produced and consumed continuously in real time. The phenomenon is most commonly known as streaming data. Traditionally, software for spatial analysis, such as a Geographical Information System (GIS) or spatial database, was created and optimized for the batch processing ofdata. However, the inherent characteristics of streaming data provide new challenges for data-stream processing systems, which have not yet been solved. In this paper, we propose enhancing systems for the handling and analysis of streaming data through the use of spatial operators. We identify Complex Event Processing (CEP) as a promising underlying concept for such a system and use the (open source) self-service IoT toolbox ‘StreamPipes’ as a representative for this. On the basis of a review of the literature, we selected 6 core types of spatial operator and implemented 33 basic spatial operators in 11 groups. These can be combined with the existing non-spatial operators for in-depth analysis of streaming data that involves spatial dimensions.
... While our analysis is based on a simulation, the same technology could be used in a real (prototypical) system if supplied with data from a computer vision subsystem and integrated into a web application. As part of our work on location based need matching systems (Bucher et al. 2017), we are working on different location proof system implementations. ...
Chapter
Full-text available
A large number of online rating and review platforms allow users to exchange their experiences with products and locations. These platforms need to implement appropriate mechanisms to counter malicious content, such as contributions which aim at either wrongly accrediting or discrediting some product or location. For ratings and reviews of locations, the aim of such a mechanism is to ensure that a user actually was at said location, and did not simply post a review from another, arbitrary location. Existing solutions usually require a costly infrastructure, need proof witnesses to be co-located with users, or suggest schemes such as users taking pictures of themselves at the location of interest. This paper introduces a method for location proofs based on visual features and image recognition, which is cheap to implement yet provides a high degree of security and tamper-resistance without placing a large burden on the user.
Conference Paper
Full-text available
The aim of this study was to analyze the 2001 Excite query log to investigate the extent and variation of Web queries containing geographic terms. In particular, an investigation into what people search for when they use geographic terms, the ways in which they describe a geographic location, the terminology used to find geographically related information and the structure of users’ queries when looking for geographically related information on the Web. This study also attempted to determine how geographically related queries differ from other queries. Geographically related queries formed nearly one fifth of all queries submitted to Excite, the terms occurring most frequently being place names. Geographic queries were also shown to be longer than average and the association of two or more terms within geographic queries was found to be high.
Article
Full-text available
Sustainable Mobility is a key goal to reduce pollution and improve quality of life. Information Technology can highly support Sustainable Mobility in many ways, such as highlighting to commuters the most suited means of transport for a given trip, or optimizing the vehicle’s occupancy rates by dynamically arranging carpools. In this paper, we present a solution for a Cloud computing-based platform, meant to improve Sustainable Mobility. In particular, the proposed system helps users in choosing a transport solution according to its ecological footprint, matching his/her needs, preferences, and actual location. Moreover, it exploits a geosocial network to improve the users’ confidence in the rides arranged with other passengers. From a technological point of view, the platform takes advantage of the Cloud infrastructure, resulting in a Software plus Services solution. In particular, the route calculation algorithm is based on a composition of different services available in the Cloud.
Article
Full-text available
Private transport accounts for a large amount of total CO2 emissions, thus significantly contributing to global warming. Tools that actively support people in engaging in a more sustainable life-style without restricting their mobility are urgently needed. How can location-aware information and communication technology (ICT) enable novel interactive and participatory approaches that help people in becoming more sustainable? In this survey paper, we discuss the different aspects of this challenge from a technological and cognitive engineering perspective, based on an overview of the main information processes that may influence mobility behavior. We review the state-of-the-art of research with respect to various ways of influencing mobility behavior (e.g., through providing real-time, user-specific, and location-based feedback) and suggest a corresponding research agenda. We conclude that future research has to focus on reflecting individual goals in providing personal feedback and recommendations that take into account different motivational stages. In addition, a long-term and large-scale empirical evaluation of such tools is necessary.
Article
Full-text available
Private transport accounts for a large amount of total CO2 emissions, thus significantly contributing to global warming. Tools that actively support people in engaging in a more sustainable life-style without restricting their mobility are urgently needed. How can location-aware information and communication technology (ICT) enable novel interactive and participatory approaches that help people in becoming more sustainable? In this survey paper, we discuss the different aspects of this challenge from a technological and cognitive engineering perspective, based on an overview of the main information processes that may influence mobility behavior. We review the state-of-the-art of research with respect to various ways of influencing mobility behavior (e.g., through providing real-time, user-specific, and location-based feedback) and suggest a corresponding research agenda. We conclude that future research has to focus on reflecting individual goals in providing personal feedback and recommendations that take into account different motivational stages. In addition, a long-term and large-scale empirical evaluation of such tools is necessary.
Conference Paper
Full-text available
We outline a system that is able to publish and match complementary spatio-temporal needs of people, e.g., the need for carpooling. Key points discussed are the modeling and publishing of needs, their specification by the user, and the efficient processing of match queries.
Article
Transportation systems are facing unprecedented challenges in the 21st century. Increasing the efficiency of transportation systems alone will not solve these problems and may exacerbate them. Instead, we must extract new transportation capabilities related to more cooperative decision-making across a wide range of time horizons, spatial scales and decision contexts. This paper discusses the role of sensed transportation, geographic information science and social media to cultivate transportation systems where participants share, cooperate and act collectively to solve operational, tactical and strategic mobility and accessibility problems. This paper also provides a vision of the future by imaging a seamless multimodal transportation system combined with a virtual environment where data streams are fused, interpreted and made available with tools for human engagement and shared decision making. This paper concludes by outlining a GIScience-centric research agenda.
Conference Paper
We argue that current technical and legal attempts aimed at protecting Geoprivacy are insufficient. We propose a novel 2-dimensional model of privacy, which we term "civilized cyberspace". On one dimension there are engineering, social and legal tools while on the other there are different kinds of interaction with information. We argue why such a civilized cyberspace protects privacy without sacrificing personal freedom on the one hand and opportunities for businesses on the other. We also discuss its realization and propose a technology stack including a permission service for geoprocessing.
Article
Context-awareness in location-based service studies has attracted more and more attention. Awareness of a user's location and the users themselves has become the centre of these services. This paper discusses a survey conducted for developing a personalised location-based property hunting system (LBSPH), and a personalised LBSPH architecture developed from the survey results. What users wanted in the way of essential property and neighbourhood information, the best way to display the information, and the common use cases of such a location-based service (LBS) were identified from the survey. The LBSPH architecture was designed to reflect these preferences.
Conference Paper
Linked Data is an RDF-based transition from the document oriented Web into the Semantic Web, and the amount of data published as linked data steadily. Hence, The RDF graph matching algorithm becomes the technical foundation of many tasks in Semantic Web, such as semantic search, data fusion, ontology matching, data filter and dissemination. An RDF-graph-matching-based query can enables searching with additional semantic information, so that it can be utilized for obtaining expected ranking in semantic search and personalized information retrieval on the web of data. Yet, the need for this approach is disregarded. This paper proposes a novel similarity-oriented RDF graph matching approach for ranking linked data, which considering the element-level and structure-level similarity of statements, and also the similarity of URIs and blank nodes in RDF graphs. The efficiency of this approach is improved over the traditional RDF graph matching algorithm. And the effectiveness is improved by analyzing and measuring the structure-level similarity of statements. The experimental results shows that this approach can effectively measure the similarity between RDF graphs, and also returns results with respect to a query RDF graph as a ranked set of promising alternatives.
Conference Paper
Traffic congestion is a serious problem in many urban areas around the world. It causes a multitude of negative impacts from producing air pollution and carbon dioxide emissions, to wasting fuel and increasing travel time, among many others. Carpooling is one of the most effective solutions to traffic congestion, and involves decreasing the number of cars on a roadway by increasing vehicle occupancy. In this paper, we propose an intelligent carpool system called BlueNet. The proposed system provides a real-time carpool service by involving two significant proposed modules: the Mobile Client module and the Cloud Global Carpool Services module. In our proposed Mobile Client module, users can use their smartphones or other devices to find carpool matches by utilizing the proposed Cloud Global Carpool Services module via the mobile communication network. The following Cloud Global Carpool Services module is used to deal with carpool requests and generate suitable carpool match results through the proposed real-time carpool-matching algorithm. To develop a carpool service that can be used around the world, we integrated the open GIS system with abundant global geographical information into the proposed Cloud Global Carpool Services module. The experimental results show that the proposed real-time carpool matching algorithm dramatically reduces the processing time required to generate the carpool matching results. This is accomplished by dynamically adjusting the random ratio value.