Smart Space-based Ridesharing Service in e-Tourism Application for

Karelia Region Accessibility

Ontology-based Approach and Implementation

Alexander Smirnov

, Nikolay Shilov

, Alexey Kashevnik

, Nikolay Teslya

and Santa Laizane

St. Petersburg Institute for Informatics and Automation RAS (SPIIRAS), St. Petersburg, Russia

Center for Internet Excellence, University of Oulu, Oulu, Finland

Keywords: Ridesharing, Ontology, Smart-M3, Smart Space, Smart Museum, Questionnaire.

Abstract: The paper describes a ridesharing service proposed for improving tourism accessibility in Russian Karelia

region. The ridesharing service has been developed using Smart-M3 information sharing platform as a smart

space infrastructure, which increases the service scalability, stability and speed, as well as reduces the

network load. The presented service is the first implementation of ridesharing concept based on the Smart-

M3 platform. The paper describes technical studies carried out to develop the prototype of ridesharing

service as well as empirical studies. An effective matching algorithm, which finds correspondences between

driver paths and tourist start and end points, and two heuristics significantly reducing path matching time,

have been presented. Besides, data analysis of online questionnaire developed to better cognize whether

prospective customers will accept ridesharing services as an alternate mode of transportation is discussed.

1 INTRODUCTION

The unique cultural and historical heritage, abundant

nature and landscape of the Russian Karelia region

provide great opportunities for tourism development

in the region. There are more than four thousand

cultural, historic and nature objects, for instance

eminent Kizhi ensemble, Vaalam Island and

Monastery, Solovetsky Islands and Monastery.

However, to facilitate the region tourism

development appropriate infrastructure should be

deployed. For instance, the lack of convenient public

transportation affects the accessibility of attractive

tourism destinations. Besides, the use of available

transportation alternatives, for example taxi, results

in considerable costs for tourist to reach the places

of interest. Thus, to improve tourism object

accessibility in the Karelia region a prototype of

ridesharing service has been developed.

Ridesharing, also known as carpooling, lift-

sharing or covoiturage, is a shared use of a car by

the driver and one or more passengers, usually for

commuting (Abrahamse and Keall, 2012).

Ridesharing enables travellers to save travel costs

and provides an alternative transportation way for

tourists to reach desired destinations. Moreover,

ridesharing is an eco-friendly and sustainable

transportation technology (Cho et al., 2012) and

therefore has been proposed, for instance, as a

promising way to reduce carbon emissions.

Recently, there has been growing interest in the

use of the web and computing methods in assisting

with ridesharing (Kamar and Horvitz, 2009). There

is a number of possibilities enabling search of fellow

travellers: public forums and communities (like,

eRideShare (eRideShare.com), PickupPal

(PickupPal.com), Zimride (Zimride.com),

RideshareOnline (RideshareOnline.com),

rideshare.511.org (Rideshare. 511.org), CarJungle

(CarJungle.ru), Podorozhniki (Prodorozhniki.com);

private Web-services (e.g., Zimride service provides

a private interface for universities and companies);

and mobile applications (e.g., PickupPal, Avego

(Avego.com).

However, most of mentioned services provide

platforms for finding fellow travelers offline, rather

than real time mobile services for generate rideshare

plans. In everyday life people often make

spontaneous decisions and are keener to pursue own

schedules. Thus, the service should enable dynamic

formation of carpools depending on current situation

and people preferences. In this paper, dynamic

ridesharing service implementation for mobile

devices is described.

591

Smirnov A., Shilov N., Kashevnik A., Teslya N. and Laizane S..

Smart Space-based Ridesharing Service in e-Tourism Application for Karelia Region Accessibility - Ontology-based Approach and Implementation.

DOI: 10.5220/0004419005910598

In Proceedings of the 8th International Joint Conference on Software Technologies (ICSOFT-PT-2013), pages 591-598

ISBN: 978-989-8565-68-6

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

Figure 1: System working scenario.

Majority of existing ridesharing services for

mobile devices, for instance PickupPal.com,

Avego.com uses the client-server architecture what

affects service scalability. Hence, to increase the

service scalability, stability and speed, as well as to

reduce network load, the prototype of the

ridesharing service has been developed using the

decentralized smart space infrastructure. Use of open

sourced Smart-M3 information sharing platform

(Honkola et al., 2010) makes it possible to

significantly simplify further development of the

system, include new information sources and

services, and makes the system highly scalable.

Presented service is the first implementation of

ridesharing concept based on the Smart-M3

platform.

The aim of this paper is to describe technical

studies carried out to develop the prototype of

ridesharing service as well as empirical studies what

forms encouraging ground for further service

development.

The system working scenario can be found in

Section 2. Section 3 introduces the logistic service

ontology used to enable interoperability between

different devices in the smart space. An effective

matching algorithm, which finds correspondences

between driver paths and tourist start and end points,

and two heuristics significantly reducing path

matching time, have been presented in Section 4.

There are technical studies, focusing on the

development of ridesharing support systems with

travel route matching techniques (Jin and Hu, 2012);

(Cho et al., 2012). The main difference of the

proposed approach is that there are two heuristics

developed, which significantly reduce matching

time. In addition, a data analysis of online

questionnaire developed to better cognize whether

prospective customers will accept ridesharing

services as an alternate mode of transportation, is

provided in Section 5. Main results are summarized

in Conclusion.

2 RIDESHARING SERVICE

SCENARIO

Common service working scenario is shown in

Figure 1. A tourist fills information about his/her

schedule, preferences (Figure 2, a), most frequent

routes, additional constraints, for example, max.

delay, max. detour, social interests, etc. (Figure 2, b)

using ridesharing service mobile application. Then,

internal processing and depersonalization of

provided information is implemented and it is

transferred into the smart space.

After, using the algorithm for finding matching

driver and passenger paths, the groups of fellow

travelers are formed. Finally, within short period of

time information about possible fellow travelers,

their profiles, meeting points, meeting time, full

recommendations about the route are provided to the

tourist (Figure 3, a-d).

Figure 2: User's routes and preferences definition.

Smart Space

Drivers

Tourists

The interaction via

the GUI

Mobile device

List of

possible tourists

The interaction via

the GUI

Transferring

information about the

routes of the users

List of

possible drivers

Mobile device

ICSOFT2013-8thInternationalJointConferenceonSoftwareTechnologies

592

a) Driver's path without ridesharing b) Tourist's path without ridesharing

c) Driver's path with ridesharing d) Tourist's path with ridesharing

Figure 3: Found routes.

Figure 4 presents an example of preliminarily

museum attending plan using ridesharing service.

Let´s assume that a tourist is going to

Petrozavodsk and he/she would like to attend

National Museum, Industry Museum, Local History

Museum, Polar Odysseus Museum, and Geological

Museum. Figure 4 presents the preliminary attending

plan for a tourist. The tourist needs to specify in

ridesharing service next place, which he/she would

like to attend and the service automatically finds

drivers, which will pick-up a tourist and drive

him/her to the place near the specified museum.

3 Smart-M3 PLATFORM

The open source Smart-M3 platform (Smart-M3 at

Sourceforge) has been used for the pilot

implementation of the presented ridesharing service.

Usage of this platform makes it possible to

significantly simplify further development of the

system, include new information sources and

services (Figure 5), and makes the system highly

scalable. The key idea of this platform is that the

formed smart space is device, domain, and vendor

independent. Smart-M3 assumes that devices and

software entities can publish their embedded

information for other devices and software entities

through simple, shared information brokers.

Information exchange in the smart space is

implemented via HTTP using Uniform Resource

Identifier (URI) (Berners-Lee et al., 2005). Semantic

Web technologies have been applied for

decentralization purposes. In particular, ontologies

are used to provide for semantic interoperability.

Figure 4: An example of museum attending plan in the

capital city (Petrozavodsk) of the Republic of Karelia.

4 THE LOGISTIC SERVICE

ONTOLOGY

The logistic service ontology describes the domain

area of ridesharing at the macro level (Figure 6).

The macro level ontology is based on integration

of parts of the mobile devices’ ontologies. The

logistics service ontology consists of three main

parts: "actors", "vehicles", and "paths".

The "actors" are: "drivers" and "passengers". All

of them are associated with class "vehicles" and

have "paths". For example, "driver" has his/her own

car and several points defining his/her home, work

and other locations. "Passenger" may prefer some

vehicle type and has points of home, work, and other

locations.

The classes "driver" and "passenger" are

subclasses of the class "actor" and inherits all its

properties with two own properties. The class

"passenger" is a subclass of the class "actor" and

Polar Odysseus

Museum

Geological

Museum

Start

Local History

Museum

Industry Museum

ational

Museum

SmartSpace-basedRidesharingServiceine-TourismApplicationforKareliaRegionAccessibility-Ontology-based

ApproachandImplementation

593

Figure 5: Smart-M3 platform.

inherits all its properties with own property "detour"

the same as in the class "driver".

For the path definition, the set of points is used.

This set is an ordered list of key points obtained as

result of the shortest path searching algorithm (e.g.,

Dijkstra or A*).

More details about the logistics service ontology

can be found in (Smirnov et al., 2010).

Figure 6: Logistics service ontology on the macro level.

5 ALGORITHM FOR MATCHING

DRIVER AND PASSENGER

PATHS

The problem of finding a matching path between the

driver and the passenger in the ridesharing service

can be formulated as follows: it is needed to

determine the possibility of ridesharing between

users, based on the information about their routes

and restrictions set by users’ services. The following

algorithm describes the procedure of finding a

matching path acceptable for the driver and the

passenger in the presented ridesharing service.

Let A be the start point and B be the end point of

the pedestrian's path. C is the start point and D is the

end point of the driver's path. The shortest driver's

path, is indicated by the solid line in Figure 7 (in

generally, CD is not a straight line; it depends on the

map of the region). The driver and pedestrian move

almost in the same direction and in some parts of the

routes the driver can give the pedestrian a ride. This

situation is indicated in the figure by the dotted line

(the CABD path) and it is the simplest situation,

because the meeting points match with the start and

end points of the pedestrian’s path. A more difficult

situation is searching for a meeting point when it

belongs neither to the driver’s shortest path nor to

the pedestrian’s one, but satisfies both the driver and

the passenger. One of the possible situations is

indicated in the figure by the dash-dot line with the

meeting points E and F (the CEFD path).

Figure 7: Matching driver and passenger paths.

Thing

Vehicle

Actor

Path

Car

Family Car

Bus

Point

Driver

Passenger

is-a

is-

is-a

role

part-o

associated wit

has

ICSOFT2013-8thInternationalJointConferenceonSoftwareTechnologies

594

The general scheme of the matching route searching algorithm will be follows:

FOR EACH driver DO

FOR EACH passenger Do

Find_mathing_path(driver.path,passenger.path); // according to the above

scheme

constraint_checking();

IF ALL constraints IS performed THEN

set_passenger_for_driver();

ENDFOR;

These points have to meet the following restrictions:

1. The distance between the start point of the

passenger and his/her meeting point should be

less than the maximum allowed detour of the

passenger (dotted circle around point A).

2. The distance between the end point of the

passenger and his/her drop-off point should be

less than the maximum allowed detour of the

passenger (dotted circle around point B).

3. The driver's detour should be less than the

maximum allowed detour.

The goal functions for finding the meeting points

are:

─ Shortest total path (interesting for the driver);

─ Minimal waiting time (interesting for the driver

and passenger);

─ Shortest distance between the passenger’s start

and end points and meeting points (interesting

for the passenger).

As a result, the general task of matching paths has

the exponential complexity; therefore, it is necessary

to apply heuristics to reduce the task dimension.

There are two heuristics proposed to reduce the

algorithm complexity. The first one allows roughly

reduce search space in short time and after that the

second heuristic allows to reduce search space even

more. However, computation complexity of a

second heuristic is higher than first; therefore, it has

to be used after the first heuristic.

5.1 Heuristics 1

Assumption: There is no need to calculate matching

paths for all pairs of drivers and passengers. It is

enough to build a set of candidate passengers for

every driver.

















































(1)







 































(2)

where pp

, pp

— the start and the end points of the

passenger's path, dp

— driver’s path point i,

PDetour, DDetour — detours of the driver and the

passenger.

5.2 Heuristics 2

Assumption: There is no need to search through all

possible combinations of meeting points. The

following alternative sub-heuristics help to reduce

the number of the possible combinations.

The first sub-heuristics selects points of the

sector from which the driver starts. Figure 8 shows

the situation when there is only one point ("C" point)

meeting constraints (1) and (2). To determine the

potential meeting points it is needed to calculate the

angle (3) and select points in the area:

 





,





 (points L and M in Figure 8).

arctg























(3)

Figure 8: The first sub-heuristics.

SmartSpace-basedRidesharingServiceine-TourismApplicationforKareliaRegionAccessibility-Ontology-based

ApproachandImplementation

595

Figure 9: The first sub-heuristics with two driver's points.

Point A will always be within the list of the

possible points as the passenger’s start or end point.

If there are more than one point meeting constraints

(1) and (2), then the search area expands. This

situation is shown in Figure 9 with two points C and

F meeting the constraints (1) and (2), and point N is

also included in the expanded area.

The negative sides of this sub-heuristics are:

─ selected points can be further than the driver’s

maximal detour;

─ some of potential meeting points can be lost if an

incorrect angle is chosen.

The second sub-heuristics (Figure 10) selects

meeting points in the intersections of the circles of

radius PDetour around the passenger’s start and end

points with the circles of radius DDetour around the

points of the driver’s path.

In this case, all of the selected points are

potentially reachable for both the driver and the

passenger, with no need to determine the angle that

restricts the selection area. The selection area can be

expanded via increasing the number of the driver's

path points meeting constraints (1) and (2).

Both sub-heuristics require the following

constraints to work effectively:

─ A large amount of drivers. Heuristics have strong

limitations and filter out a lot of points. If there

are no enough drivers, then the use of the

heuristics will rarely produce positive result.

─ A small value of DDetour. Heuristics will not be

helpful with a large value of DDetour.

─ Uniform distribution of roads on the map. The

uneven distribution of roads (rivers, lakes, etc)

leads to a lack of roads in some sectors, which

could lead to the loss of possible meeting points

due to the need to detour around the obstacles

and to pick up the pedestrian on the other side.

─ Both heuristics 1 and heuristics 2 are used in the

logistics service prototype. Without using the

heuristics, the system finds from 10 to 12

meeting points for each pair of driver and

passenger and it needs to check all of 100–144

combinations to find the best one. With using the

heuristics, the number of points is reduced to 8-9

points with 64–81 combinations for each pair of

driver and passenger.

6 EVALUATION

The algorithm for matching driver and passenger

paths has been tested using the following computer:

Intel Pentium 4, 1.6 GHz, RAM: DDR1 512 Мб. For

the experiments, the algorithm with proposed

heuristics has been run on random datasets as input

parameters for the predefined drivers and passengers

(see, Table 1). Each dataset includes coordinates of

start and end points of a fellow traveler. Experiments

show that heuristics help to reduce the time of search

in more than 1.5 times.

Figure 10: The second sub-heuristics.

Table 1: Results of the experiments.

Drivers Passengers Matching time, sec

1 1 0,0135

5 5 0,0316

10 10 0,0641

20 20 0,2248

40 40 1,5462

60 60 2,2416

80 80 3,4725

Figure 11 presents relationship between

matching time and number of drivers and passengers

in the service.

ICSOFT2013-8thInternationalJointConferenceonSoftwareTechnologies

596

Figure 11: Relationship between matching time and

number of drivers and passengers in the service.

7 QUESTIONNAIRE-BASED

ANALISIS

To better cognize, whether our prospective

customers will accept ridesharing services as an

alternate mode of transportation, an online

questionnaire has been developed.

7.1 Questionnaire Design

and Respondent Selection

The questionnaire was constructed using Likert

Scale statements and open-ended questions. The

statements in Likert Scale were structured in a way

that there were five choices i.e. (Strongly disagree;

Disagree; Not Sure; Agree; Strongly Agree). We

opted for Likert Scale when designing the

questionnaire as it gives participants an opportunity

to specify their responses on different levels. In

response, the participants expressed their opinions

pertaining to costs associated with travelling,

security concerns (for instance, travelling with

strangers), and issues relating to travel

schedules/routes etc. Open-ended questions were

poised to gain richer insight into peoples’ attitudes

towards ridesharing. For example, to establish main

reasons why people use the service or why they

make specific choices while using ridesharing

service abroad.

Ridesharing service has been proposed as an

assuring approach to improve tourism in Russian

Karelia accessibility for local tourists and those who

travel across the border. The study took place during

November and December 2012 and responses were

gathered both from Russian and Finnish population.

In all, forty eight (N=48) responses were collected.

However, forty six (N=46) were used to perform the

analysis because two responses were incomplete.

Out of the 46 respondents, there were 33 males

(71.7%) and 13 females (28.3%) with an average

age of 32 years (ranging from 20 to 68 years). From

Finland (primarily Oulu region), 26 responses were

received and 26 participants completed the

questionnaire from Russia (St. Petersburg and

Petrozavodsk).

7.2 Analysis of Questionnaire Results

More than half of the participants (56.5%) reported

that they had never used ridesharing in their

respective countries. Interestingly, a bigger number

of participants (76.1%) stated that they did not use

ridesharing abroad. A significant number of

participants (32.6%) stated that they rarely used

ridesharing and the remaining (10.8%) informed that

they either used it on monthly, weekly or daily basis.

More than half of the respondents (54.3%)

confirmed that they would prefer to use the mobile-

based ridesharing application. The responses suggest

that people have an inclination to use the service for

reasons such as reduced travel costs (80.7%) and

saving time (69.5%). These findings were confirmed

by the responses to open-ended questions where

majority of the respondents stated that they preferred

ridesharing to save travel costs and time. Yet, a

number of participants (39.1%) stated that they

would use ridesharing only when it is time efficient

when compared with public transport, or, when their

destination is not accessible via public transport

(36.9%). Besides, there is a similar amount of

respondents who stated that they will use ridesharing

even though a destination would be accessible by

public transportation (45,6%) and also when by

public transportation it would be faster (32,6%).

Therefore, we can cautiously propose that

ridesharing is a promising solution to improve

tourism object accessibility.

Although the data analysis indicates positive

attitudes towards the acceptance and potential usage

of ridesharing, the service will need to be

incorporated with features that will enhance security.

Security aspects were stated as main reasons why

people would not use ridesharing. For instance, the

majority of the respondents (65.2%) will use

ridesharing only together with people who is

suggested by personally known person. Therefore

developing ridesharing further is a prerequisite

through social networks, facilitation of comments

and feedback and an option to share individual

experiences.

SmartSpace-basedRidesharingServiceine-TourismApplicationforKareliaRegionAccessibility-Ontology-based

ApproachandImplementation

597

8 CONCLUSIONS

The paper proposes the smart space-based

ridesharing service to improve tourism in Russian

Karelia accessibility. The paper describes the service

architecture, main algorithms, used heuristics (which

help to reduce the time of search in more than 1.5

times), implementation, and the questionnaire

developed to cognize whether prospective tourists

will accept ridesharing services as an alternate mode

of transportation. Developed questionnaire analysis

shows applicability of this type of transportation in

Karelia region (respondents would like to use the

ridesharing service for reduced travel costs (80.7%)

and saving time (69.5%)). Smart-M3 information

platform is used as a smart space infrastructure for the

presented approach. Use of this platform makes it

possible to significantly increase the scalability and

extensibility of the prototype system. The algorithm

for finding appropriate fellow travelers for drivers as

well as definition of acceptable pick-up and drop-off

points for them is presented in the paper. This

algorithm can effectively find appropriate fellow

travelers for drivers. Importance of driver path and

passenger start and end points matching has been

confirmed by analysis of questionnaire results. For

instance, majority of the respondents would

rideshare if there is no need to change their travel

routes (50%) and when it fits well with their

schedules (73.9%).

Although there are positive attitudes towards the

acceptance and potential usage of ridesharing, the

service will need to be incorporated with features

that will enhance security. Besides, integration of

social networks, facilitation of comments and

feedback and an option to share individual

experiences are planned in the future work.

ACKNOWLEDGEMENTS

This research is a part of grant KA322

«Development of cross-border e-tourism framework

for the programme region (Smart e-Tourism)» of

Karelia ENPI programme, which is co-funded by the

European Union, the Russian Federation and the

Republic of Finland. The presented results are also a

part of the research carried out within the project

funded by grant # 13-07-00336-a of the Russian

Foundation for Basic Research.

REFERENCES

Abrahamse, W., Keall, M., 2012. Effectiveness of a web-

based intervention to encourage carpooling to work: A

case study of Wellington, New Zealand. In: Transport

Policy, Vol. 21, pp. 45–51.

Kamar, E., Horvitz, E., 2009. Collaboration and Shared

Plans in the Open World: Studies of Ridesharing,

Microsoft Research Technical report.

eRideShare.com. URL: http://erideshare.com, last access

date 15.04.2013.

PickupPal. URL: http://www.pickuppal.

com, last access date 15.04.2013.

Zimride. URL: http://www.zimride.com, last access date

15.04.2013.

RideshareOnline.comRideshareOnline. URL:

http://www.rideshareonline.com, last access date

15.04.2013.

Rideshare 511. URL: http://rideshare.511.org, last access

date 15.04.2013.

CarJungle. URL: http://www.carjungle.ru, last access date

15.04.2013.

Podorozhniki.com, URL: http:// http://podorozhniki.com/,

last access date 15.04.2013.

Avego. URL: http://www.avego.com, last access date

15.04.2013.

Jin, B., Hu, J., 2012. Towards scalable processing for a

large-scale ride sharing service. In 9-th International

Conference on Ubiquitous Intelligence and Computing

and 9

International Conference on Automatic and

Trusted Computing.

Cho, A., Yasar, A., Knapen L., Bellemans, T., Janssens,

D., Wets, G., 2012. A conceptual design of an agent-

based interaction model for the carpooling application.

In the 1-st Internationsl Workshop on Agent-based

Mobility, Traffic and Transportation Models,

Methodologies and Applications. Elsevier Ltd.

Smart-M3 at Sourceforge, 2012. URL:

http://sourceforge.net/projects/smart-m3, last access

date 15.04.2013.

Berners-Lee, T., Fielding, R., Masinter, L., 2005. RFC

3986 – Uniform Resource Identifier (URI): Generic

Syntax, URL: http://tools.ietf.org/html/rfc3986.

Honkola, J., Laine, H., Brown, R., Tyrkko, O. 2010.

Smart-M3 Information Sharing Platform. In: Proc.

IEEE Symp. Computers and Communications. IEEE

Comp. Soc.; pp. 1041-1046.

Smirnov, A., Kashevnik, A., Shilov, N., Paloheimo, H.,

Waris, H., Balandin, S., 2010. Smart Space-Driven

Sustainable Logistics: Ontology and Major

Components, Sergey Balandin, Andrei Ovchinnikov

(eds.). In: Proceedings of the 8th Conference of Open

Innovations Framework Program FRUCT,

Lappeenranta, Finland, 9-12 Nov., 2010, pp. 184-194.

ICSOFT2013-8thInternationalJointConferenceonSoftwareTechnologies

598