Representing GeoData for Tourism with Schema.org

Oleksandra Panasiuk, Zaenal Akbar, Thibault Gerrier and Dieter Fensel

University of Innsbruck, Technikerstrasse 21a, Innsbruck 6020, Austria

Keywords:

Semantic Annotation, Geographical Data, schema.org, e-tourism.

Abstract:

A large amount of tourism data on the web, representing different touristic services, refers to information

which is geographically located. With the intensive development of artiﬁcial intelligence, interest in the an-

notation of data is continuously increasing. It is therefore important to describe all tourist needs. To be

understandable to search engines, chatbots or other personal assistant systems, content data should be struc-

tured, well-formed and semantically consistent. Schema.org is a de-facto standard for marking up structured

data on the web. In this paper we show how to annotate geographical information related to different touristic

services and activities (e.g. hotels, restaurants, events, hiking and climbing trails) available on an interactive

map using the schema.org vocabulary.

1 INTRODUCTION

When visiting tourist destinations, especially new or

big ones, tourists are typically challenged to decide

the best plan on how to visit a number of points of in-

terest with limited resources such as time and budget.

This challenge can be seen as a tourist trip design pro-

blem (Vansteenwegen and Van Oudheusden, 2007),

where the best route can be determined by combining

user interests, attraction, and trip information in re-

spect to time and budget constraints. Different expert

systems for trip planning has been developed to over-

come the challenge, for example for a city trip (Van-

steenwegen et al., 2011) and multimodal tour (Gava-

las et al., 2015). To make decisions, those systems

built a knowledge base where points of interest’s data

were obtained from tourists ofﬁces, tourist portals, or

Wikipedia.

Various tourism related information can be found

at numerous information sources including geo-

graphical data which are usually presented on maps.

These data could be represented differently among sy-

stems, causing problems for data integration or data

linking. A way to overcome this data representation

disparity is through data annotation. Annotation can

be seen as a process of connecting two pieces infor-

mation, typically with a purpose to give a better expla-

nation of the data. Speciﬁcally for semantic annota-

tion, the additional information is intended to connect

data with its “meaning”, in a way so that it can be re-

presented and accessed uniformly. Data that are avai-

lable in distributed sources can be read and proces-

sed by machine automatically. Semantic annotation

can be performed manually or (semi)-automatically

for different purposes, for example to annotate digi-

tal music in such a way that a music item can be lo-

cated easily according to users choice (Rahman and

Siddiqi, 2012), improving search accuracy for web se-

arches (Fuentes-Lorenzo et al., 2013), including geo-

graphical data (e.g. points of interest) on a map (Ruta

et al., 2012; Stadler et al., 2012).

Semantically annotated content impacts the tou-

rism industry signiﬁcantly. It feeds search engines

structured data in a way that can be presented in a

more interesting way visually, for example Google

Search has rich features where events will be presen-

ted in a structured layout, or recipes in a carousel

It increases online visibility of a typical hotel’s web-

site by up to 20% (Fensel et al., 2016). The annota-

ted content can be consumed by intelligent applica-

tions, for example to guide a semi-automatic distri-

bution of content to multiple online communication

channels (Akbar et al., 2014). Latest technologies

such as chatbots and personal digital assistants pro-

cess annotated content to answer users requests in an

automatic fashion.

In this paper, we present our work to annotate geo-

data (geographical data) which are related to the tou-

ristic domain in the region of Tyrol, Austria by using

the Schema.org vocabulary. Our intention is clear, an-

notating geo-data information will not only increase

https://developers.google.com/search/docs/guides/

search-features

Panasiuk, O., Akbar, Z., Gerrier, T. and Fensel, D.

Representing GeoData for Tourism with Schema.org.

DOI: 10.5220/0006755102390246

In Proceedings of the 4th International Conference on Geographical Information Systems Theory, Applications and Management (GISTAM 2018), pages 239-246

ISBN: 978-989-758-294-3

239

the region’s online visibility but also enable intelli-

gent applications to consume it. We did not use any

vocabulary that was developed speciﬁcally for geo-

data, because most of this vocabulary did not receive

sufﬁcient adoption. Instead, we use more common

vocabulary that has been widely adopted where an ex-

tension will be proposed whenever necessary. To be

precise, our contributions are: (i) domain speciﬁca-

tion for geo-data, (ii) an approach to integrate anno-

tation to a website, (iii) a signiﬁcant amount of an-

notation of geo-data of the Tyrol region. The paper

will be organized as follow, Section 2 lists a few ex-

isting related works and discusses our contributions.

Section 3 presents our analysis and deﬁned speciﬁ-

cation for geo-data, followed by our method to cre-

ate annotations based on the deﬁned speciﬁcation in

Section 4. Sections 5 and 6 discuss our results and

their evaluation, and ﬁnally we conclude our work and

list some potential future works in Section 7.

2 RELATED WORKS AND

MOTIVATION

In this section, we outline our main motivation for an-

notating geographical data as well as a few existing

related works. Touristic industry can not be separated

from the development of the region of Tyrol, Austria.

In 2015/2016, 11.5 million guests were visited the re-

gion, produced 47.6 million overnight stays

A way to

help the industry is by utilizing semantic web techno-

logies, including semantic annotations (Akbar et al.,

2017).

2.1 Motivation

Intelligent applications require a system that feeds

them with knowledges in order to make decisions. A

widely used system known as Knowledge Graph

, po-

pularized by Google, uses a graph database to pro-

vide structured and detailed information integrated

from distributed sources. It has been used in various

sectors, for example for sales assistant (Kim, 2017)

and question answering system (Kumar et al., 2017).

Knowledge graph can be built from semantically an-

notated content, and therefore it is necessary to have

a large amount of annotation to produce a complete

knowledge. According to the Web Data Commons

Tirol Werbung, 2016, “Der Tiroler Tourismus Zahlen,

Daten und Fakten 2016”, http:// www.tirolwerbung.at/

tiroler-tourismus/zahlen-und-fakten-zum-tiroler-tourismus/.

https://en.wikipedia.org/wiki/Knowledge Graph

http://webdatacommons.org/structureddata/

in 2016 there were around 1.24 billion HTML pages

contain structured data, it is about 38% of the whole

pages available on the Web. The number of Resource

Description Format (RDF) triples has been increased

signiﬁcantly, about 44.2 billions in 2016, compared to

24.4 billions in 2015.

Annotating geo-data with speciﬁc vocabulary is

not a new topic, there are a few existing works have

been done with variety purposes. The LinkedGeo-

Data project

transformed and published OpenStreet-

Map

geographical data using Linked Data princi-

ples

, enabling semantic-spatial search as well as geo-

data syndication (Stadler et al., 2012). Linking con-

cepts detected in books to locations (spatial informa-

tion) could produce a novel location-based recom-

mendation system (De Meester et al., 2015). And

there are a few standard exists for digital geographic

information, including ISO/TC 211

2.2 Contribution

More than just about structuring data in a uniform

representation, semantic annotations approach pre-

sents many advantages, including reasoning capabi-

lity. This reasoning capability is highly needed when

developing an intelligent application, for example to

determine whether the description of a concept is

more general than the description of another concept

(subsumption of concepts), or to ﬁnd all individuals

that are instances of a concept (retrieval of individu-

als).

There are a few existing works that have been pro-

ducing a large amount of annotation for touristic re-

lated informations (Akbar et al., 2017), but none of

them include geographical data yet. This work will

enrich those existing annotation and providing a base

for tourism related geographical data annotation.

Our work is taking advantage of the wide adop-

tion of Schema.org, where classes in this vocabulary

have been used more than 50% on average (Meusel

et al., 2015). With the support of major search engi-

nes including Google and Bing, the rate of adoption is

increasing. And therefore we are quite conﬁdent that

our approach, annotations of geographical data with

Schema.org, will also receive the same rate of adop-

tion.

http://linkedgeodata.org/

https://www.openstreetmap.org/

https://en.wikipedia.org/wiki/Linked data

https://en.wikipedia.org/wiki/ISO/TC 211

GISTAM 2018 - 4th International Conference on Geographical Information Systems Theory, Applications and Management

240

Figure 1: Example of Hiking trail.

3 METHODOLOGY OF GeoData

ANNOTATION

In this section we describe the methodology for anno-

tating the touristic services based on their geographi-

cal data and discuss the main challenges that arise”.

As a pilot we use the maps of the Tourismusverband

(TVB)

, Mayrhofen-Hippach

and Seefeld

. The

methodology consists of the following parts.

3.1 Data Content and Format Analysis

Tourist Maps are built based on the information ne-

cessary for tourist needs and related activities. That’s

why, for the annotation process, it is important to ana-

lyze the real world representation and to extract exis-

ting service types relevant in certain tourist areas ac-

cording to tourist preferences (Gretzel et al., 2004).

The next step is to analyze online representation

of touristic services displayed on the maps, data con-

tent and its format. Very often data represent different

types of information for the same touristic service,

and are semi-structured or unstructured, which cau-

ses difﬁculties for the semi-automatic annotation pro-

cess. The Interactive Maps of the TVB Mayrhofen-

Hippach and Seefeld, which we analyzed, contain a

variety of tourism-related information, including hi-

king or biking routes, accommodation, infrastructure

and so on. Besides information about geodata objects,

the maps also contains metadata information about

points on the map, for example, a contact point for a

store, the distance or altitude of a hiking route, or real-

time bus schedules at a given bus stop. The content

consists of the information from the Contwise Maps,

Destination Management Organization

https://maps.mayrhofen.at/

https://maps.seefeld.com/

an interactive map provided by General Solutions

and information from the external source Feratel

There are two main data structures in the Cont-

wise Maps: (i) Object Types (Forms) and (ii) Cate-

gories. The Object Types deﬁne the attributes of an

object. Every resource (content entry) in the Cont-

wise Maps can have only one object type.The Catego-

ries are the menu items and can be nested into a tree

structure. They are completely independent from the

object type. Resources of any object type can appear

in one or more categories and subcategories.

All this information can be accessed through a

web API which serves data in a JavaScript Object No-

tation (JSON) format.

3.2 Domain Deﬁnition

In this step we construct the set of domains which we

want to annotate. Each tourist object from the map

belongs to an object type and categories as mentioned

above and may have some subcategory. This is why

our primary task is to deﬁne domains and select pro-

per types from schema.org based on data structures in

the Contwise Maps. We extract the list with object

types, categories and subcategories and select a pro-

per type from schema.org, as shown in Table 1. The

main challenge here is that the data structures used

by General Solutions and schema.org were built for

different purposes, leading to discrepancies between

their types and properties. Some categories and ob-

ject types are too general or too detailed, some don’t

have correct English matches, and some are not cove-

red by the schema.org vocabulary.

3.3 Mapping to a Domain Speciﬁc

Subset of schema.org

From the domain deﬁnitions we get the set of different

domains. The next step is to map the data content of

these domains to schema.org in a way which suitably

represents its types and properties. The main chal-

lenge here is to ﬁnd the best and most suitable way

to map the data, i.e. to choose the right class with

properties from schema.org. For example, for a Cable

Car we choose the type CivicStructure with the follo-

wing properties: name, description, amenityFeature,

contactPoint, image, address, geo, containsPlace, po-

tentionAction,hasMap, and url. Each property has

its range, e.g. Text, Url, DateTime, QuantitativeVa-

lue.Some elements can have external properties, and

we considered them too. Geo in schema.org has the

https://general-solutions.eu

http://www.feratel.at/

Representing GeoData for Tourism with Schema.org

241

Table 1: An example of the domain deﬁnition.

Object Type Subcategory Category Schema.org type

biking - Mountainbike-Tour SportsActivityLocation

hiking Wandern Sport & Freizeit SportsActivityLocation

infrastructure Pizzeria Restaurant/ Pizzeria Restaurant

serviceProvider Ferienwohnung / Ferienwohnung / LodgingBusiness

Appartement Appartement

crosscountry Klassisch SportsActivityLocation

webcam - Video VideoObject, Place

gdi-lift - Ubersicht Sommer-Bergbahnen CivicStructure

extern-link - Skibus; Wanderbus BusStation

gdi-piste - Skigebiet SkiResort

infrastructure - Veranstaltungsort Event

infrastructure Hallenbad Baden; See; Schwimmbad PublicSwimmingPool

...

range GeoCoordinates, where GeoCoordinates is de-

ﬁned by longitude, latitude, elevation.

3.4 Domain Speciﬁcation

This section, based on previous steps, provides the

common models for annotating different touristic

domains presented on the map. For this purpose

we analyze the results from sections 3.2 and 3.3 and

choose the wide use domains. Then for each domain

we model a domain speciﬁcation, i.e. select or com-

bine the most suitable deﬁned properties and classes,

including the range types. These domain speciﬁcati-

ons give us the patterns, that can help to automate the

annotation process and make it easier (Panasiuk et al.,

2018). In Table 2 an example of SportsActivityLoca-

tion domain speciﬁcation is shown.

4 IMPLEMENTATION

In this section, we explain our implementation of the

annotation for the content Interactive Maps of Cont-

wise Maps from General Solution. The implementa-

tion consists of (i) getting data from the source, (ii)

the mapping process from the input to structured data

with the vocabulary from schema.org and (iii) annota-

tion deployment. For implementation we use Node.js,

as it is good for HTTP requests as well as manipula-

ting data in the JSON format.

4.1 Getting the Data

We get the data via General Solution API. All the

requests are simple HTTP GET requests and the re-

sults containing all of the data are in the JSON for-

mat, which has key value pairs. From the requests

we obtain an array of JSON objects for each category

and the images for each individual resource. Figure 2

shows the process of getting data via General Solution

API.

4.2 Mapping Process

The structure of JSON objects is as follows: a unique

identiﬁer, a name, a description, a ﬁeld for geographic

coordinates, an array of category ids and a ”metadata”

ﬁeld containing different information such as address,

phone number, and more speciﬁc information like the

length of a trail, if the trail is suitable for children, the

number of beds for a hotel and so on.

The ﬁrst step of the mapping consists of ﬁnding

out what schema type the resource should use based

on its categories. For that, a list is made with every

possible category id and its appropriate schema type.

Those categories were often very vague and thus were

followed by subcategories. The next step is to map

those properties that can always be mapped the same

way, those are: id, name, description, coordinates,

ﬁelds concerning address (street address, postal code,

...), images and URLS. The structure of the program is

shown in Figure 3 and can be summarized as follows:

if the resource has this or that property then map it in

this or that way. So, the program goes through all pos-

sible ﬁelds the data can have and maps it according to

our previous mappings.

The result of this mapping is an array of JSON-

LD objects using the schema.org vocabulary, where

all meaningful data from the source was annotated as

described by our mapping.

4.3 Annotation Deployment

The next step was getting those JSON-LD ﬁles into

semantify.it

. The platform has a nice API where

https://semantify.it/

GISTAM 2018 - 4th International Conference on Geographical Information Systems Theory, Applications and Management

242

Table 2: An example of the domain speciﬁcation for SportsActivityLocation.

Domain Types Set of properties Range Type Attributes

SportsActivityLocation name Text required

description Text required

image URL, ImageObject required,

multitype

hasMap Map or URL required

address PostalAddress

PostalAddress addressCountry Text, Country

name Text optional

addressLocality Text required

addressRegion Text required

postalCode Text required

streetAddress Text required

geo GeoCoordinates, GeoShape required

GeoCoordinates longitude Number required

latitude Number required

elevation Number optional

...

Figure 2: Architecture of getting data via API.

Figure 3: Architecture of mapping the data.

you can upload a few hundred JSON ﬁles at a time to a

given API-key (linked to an account or organization).

So, with just a few requests all the data is transmitted

and ready for the customer. (?)

5 EVALUATION

We evaluate our annotation model with formally va-

lidation the functional requirements of our domain

speciﬁcations with answering the competency ques-

tions (Uschold and Gruninger, 1996). We translate

questions into queries, use SPARQL and to answer

them and hence formally evaluate our ontology (Stolz

et al., 2017). We use Ontotext GraphDB

as a soft-

ware product for storing our annotation and querying

them with SPARQL.

As explained in Section 2, the semantic annota-

tions approach was utilized to not only representing

data uniformly but also to obtain reasoning capabi-

lity. Therefore, in the following SPARQL queries, we

discuss the capabilities introduced by semantic anno-

tations with Schema.org.

Question 1. Return all trails with average difﬁculty

in Seefeld region.

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

PREFIX schema:<http://schema.org/>

SELECT ?s ?name WHERE

{

GRAPH ?g

{

?s rdf:type schema:SportsActivityLocation.

?s schema:name ?name.

?s schema:amenityFeature ?feature.

?feature schema:name "Difficulty".

?feature schema:value "average".

FILTER(str(?g)="http://maps.seefeld.com")

}

According to the “SportsActivityLocation” spe-

ciﬁcation of Schema.org

, the type covers variety

more speciﬁc types such as “BowlingAlley”, “Golf-

Course”, “PublicSwimmingPool”. By using this sin-

gle query, instances of all those speciﬁc types will be

http://graphdb.ontotext.com/

http://schema.org/SportsActivityLocation

Representing GeoData for Tourism with Schema.org

243

also included in the result, in consequence of type and

sub-type relationships.

Question 2. Return all cable cars and their length in

ski resorts of Mayrhofen-Hippach map and with the

altitude of mountain station more then 1815m.

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>

PREFIX schema:<http://schema.org/>

PREFIX dbp: <http://dbpedia.org/resource/>

SELECT ?s ?name ?distance ?elevation WHERE

{

GRAPH ?g

{

?s rdf:type schema:CivicStructure.

?s schema:additionalType dbp:Cable_car.

?s schema:name ?name.

?s schema:potentialAction ?action.

?action rdf:type schema:TravelAction.

?action schema:distance ?distance.

?action schema:toLocation ?place.

?place schema:geo ?geo.

?geo schema:elevation ?elevation.

FILTER (?elevation >= 1815).

FILTER(str(?g)="http://maps.mayrhofen.at")

}

This second query requested instances of “Civi-

cStructure” which have an additional type of “Ca-

ble car” from DBpedia

. This is another advan-

tage of using semantic annotations with Schema.org,

where data can be linked with speciﬁc types from ex-

ternal vocabularies, for instance to avoid ambiguity.

6 RESULTS

In this section, we discuss results from our work of

annotating geo-data available in the maps of TVB

Mayrhofen-Hippach and Seefeld. First, we describe

the statistics of produced annotations, then discuss in

detail about entities that were annotated with multiple

types.

6.1 Annotation

In total, we produced about 58k annotations from both

maps, where an annotation could has exactly one type

or combination of several types known as multi-type

entity (MTE). Table 3 shows the statistics of produced

annotations.

As shown in this table, more than 93% of anno-

tations were produced from geo-data available in the

map of TVB Seefeld, annotated with less than 100

types available in Schema.org. Type “Event” was

dominated the annotation (above 73%), followed by

http://dbpedia.org/resource/Cable car

Table 3: Statistics of Produced Annotations.

No. Description

Maps of TVB

Mayrhofen Seefeld

1. Annotations 3.986 53.997

2. Used types 78 97

3. Annotations with MTE 1.096 2.481

Figure 4: Distribution of types usage (top 20).

“LodgingBusiness” (5%), “SportsActivityLocation”

(5%), “Restaurant” (2%), “Hotel” (2%), “BedAndB-

reakfast” (2%), and the other types were used about

less than 1%. About 22% of annotations for TVB

Mayrhofen-Hippach were used MTE and only about

4% for TVB Seefeld.

We examined further the top 20 of types used in

the maps as shown in Figure 4. The most dominant

types found are “LodgingBusiness” and “Event” in

the maps of TVB Mayrhofen-Hippach and Seefeld

respectively. Also, location for doing sport activi-

ties (annotated with type “SportsActivityLocation”)

was dominating annotations in both maps. For mo-

bility support (indicated with type “BusStation”) was

dominated by geo-data available in the map of TVB

Mayrhofen-Hippach.

6.2 Multi-types Entity Annotation

Another important results are entities that were anno-

tated with multiple types. A multi-type entity will be

required whenever the entity can be represented with

Figure 5: Annotations with multiple types.

GISTAM 2018 - 4th International Conference on Geographical Information Systems Theory, Applications and Management

244

a single type. For us, this situation leads us to iden-

tify types that are not covered by Schema.org yet, as

well as to ﬁnd types that share similar properties. As

shown in Figure 5, an entity could be annotated with

more than one type. While most of them were an-

notated with 2 types, an entity could has up to 5 or

16 types for TVB Mayrhofen-Hippach or Seefeld re-

spectively.

6.3 Discussion

We highlighted the usage of MTE in our annotations

because a point of interest on map could has variety of

organizational functions. Each function was reﬂected

by a different type. For example, a point of interest

has been annotated with 2 types, namely “ClothingS-

tore” and “ShoeStore” to represent 2 different busi-

ness functions offered by the entity. These business

functions were based on categories and sub-categories

deﬁned by the data provider, obtained through the

mapping to the deﬁned types from the selected vo-

cabulary (as explained in Section 4). Another possi-

bility to represent multi-type entities especially those

that related to e-commerce is deﬁning the entity as

“Shop” which offers various types of products such

as “Clothing”, “Bike”, “Shoe”, and so on.

7 CONCLUSION AND FUTURE

WORK

In this paper, we presented a work on representing ge-

ographical data related to touristic information avai-

lable on an interactive map using Schema.org. Our

intention was to provide clear description to all iden-

tiﬁed points of interest such that can be consumed by

intelligent application, such as search engines, per-

sonal digital assistants. Although many vocabula-

ries for geographical data speciﬁc have been propo-

sed, most of them have failed to get wide adopti-

ons. Since Schema.org vocabulary has been widely

adopted, we believe that our work will gained wider

acceptance from the Internet users. In our work we

analyzed data source and format, identiﬁed types and

categories (including sub-categories) of available ob-

jects. For every type and category, proper types from

Schema.org were identiﬁed, and domain speciﬁcati-

ons were produced. The software tool was developed

to consume the produced mapping, retrieve content

from the API, produce relevant annotations and de-

ploy them to a platform in a way ready to be consu-

med by customers. At the end, we produced about

https://maps.seefeld.com/de#resourceDetail,1653626

58k annotations for data obtained from two maps, uti-

lized about 100 different types.

Unfortunately, schema.org doesn’t offer every

possible type and property for the tourism and tra-

vel industry, so a lot of classes and features were just

mapped into a new key-value pair. For example, there

are no speciﬁc types for sport locations and activities,

such as: hiking and climbing trails, running and cy-

cling routes and ski slopes, which are so popular for

the Tyrol region. We annotate such objects with the

help of schema’s ”SportsActivityLocation”, ”Loca-

tionFeatureSpeciﬁcation”, ”TravelAction” and their

properties (e.g. amenityFeature, potentialAction, dis-

tance) to describe the duration, length, start and end

point of a trail and so on. There were also some map-

ping deﬁciencies, such as: language differences, dif-

ferent concepts and priorities for data types provided

by Feratel, General Solution and schema.org. Also,

it is worth mentioning that in some cases, a data in-

stance might inherit properties from multiple types,

known as a multi-type entity, as some objects appear

in different categories or it is required to use two types

to cover one object with schema.org.

Future work is to provide more specialized vo-

cabularies for the touristic domain to build upon the

core and start the process of extending the schema.org

standard. For this purpose the Schema Tourism Wor-

king Group

is running. Our ultimate goal is to have

all tourism related information in the region of Tyrol,

Austria semantically represented on the Web. More

than just providing a high online visibility for the re-

gion on search engines, these data also contribute to

opening new marketing channels such as chatbots and

other intelligent personal digital assistants.

ACKNOWLEDGEMENTS

This work was performed with cooperation and sup-

port from General Solutions, Tirol Werbung GmbH,

TVB Mayrhofen and Seefeld . We would like to

thank all the members of the Schema Tourism Wor-

king Group for their valuable feedback.

REFERENCES

Akbar, Z., Garc

ıa, J. M., Toma, I., and Fensel, D. (2014).

On Using Semantically-Aware Rules for Efﬁcient On-

line Communication. In Bikakis, A., Fodor, P.,

and Roman, D., editors, Lecture Notes in Computer

Science (including subseries Lecture Notes in Artiﬁ-

cial Intelligence and Lecture Notes in Bioinformatics),

https://schema-tourism.sti2.org/

Representing GeoData for Tourism with Schema.org

245

volume 8620 LNCS, pages 37–51. Springer, Prague,

Czech Republic.

Akbar, Z., K

arle, E., Panasiuk, O., S¸ims¸ek, U., Toma, I., and

Fensel, D. (2017). Complete semantics to empower

touristic service providers. In Panetto, H., Debruyne,

C., Gaaloul, W., Papazoglou, M., Paschke, A., Ar-

dagna, C. A., and Meersman, R., editors, On the Move

to Meaningful Internet Systems. OTM 2017 Conferen-

ces: Confederated International Conferences: Coo-

pIS, C&TC, and ODBASE 2017, Rhodes, Greece, Oc-

tober 23-27, 2017, Proceedings, Part II, pages 353–

370. Springer, Cham.

De Meester, B., De Nies, T., Verborgh, R., Mannens, E.,

and Van de Walle, R. (2015). Reconnecting digital

publications to the web using their spatial information.

In Proceedings of the 24th International Conference

on World Wide Web, WWW ’15 Companion, pages

749–754, New York, NY, USA. ACM.

Fensel, A., Akbar, Z., Toma, I., and Fensel, D. (2016). Brin-

ging Online Visibility to Hotels with Schema.org and

Multi-channel Communication. In Inversini, A. and

Schegg, R., editors, Information and Communication

Technologies in Tourism 2016, pages 3–16. Springer,

Bilbao, Spain.

Fuentes-Lorenzo, D., Fern

andez, N., Fisteus, J. A., and

anchez, L. (2013). Improving large-scale search en-

gines with semantic annotations. Expert Systems with

Applications, 40(6):2287–2296.

Gavalas, D., Kasapakis, V., Konstantopoulos, C., Pantziou,

G., Vathis, N., and Zaroliagis, C. (2015). The eCOM-

PASS multimodal tourist tour planner. Expert Systems

with Applications, 42(21):7303–7316.

Gretzel, U., Mitsche, N., Hwang, Y.-H., and Fesenmaier,

D. R. (2004). Tell me who you are and i will tell you

where to go: Use of travel personalities in destination

recommendation systems. Information Technology &

Tourism, 7(1):3–12.

Kim, H. (2017). Towards a sales assistant using a product

knowledge graph. Web Semantics: Science, Services

and Agents on the World Wide Web.

Kumar, A. J., Schmidt, C., and Khler, J. (2017). A know-

ledge graph based speech interface for question ans-

wering systems. Speech Communication, 92(Supple-

ment C):1 – 12.

Meusel, R., Bizer, C., and Paulheim, H. (2015). A Web-

scale Study of the Adoption and Evolution of the

schema.org Vocabulary over Time. In Proceedings of

the 5th International Conference on Web Intelligence,

Mining and Semantics - WIMS ’15, pages 1–11, New

York, New York, USA. ACM Press.

Panasiuk, O., K

arle, E., S¸ims¸ek, U., and Fensel, D. (2018).

Deﬁning tourism domains for semantic annotation of

web content. ENTER 2018 Conference on Information

and Communication Technologies in Tourism, Rese-

arch Notes in e-Review of Tourism Research, 11.

Rahman, F. and Siddiqi, J. (2012). Semantic annotation of

digital music. Journal of Computer and System Scien-

ces, 78(4):1219–1231.

Ruta, M., Scioscia, F., Ieva, S., Loseto, G., and Di Sciascio,

E. (2012). Semantic Annotation of OpenStreetMap

Points of Interest for Mobile Discovery and Naviga-

tion. In 2012 IEEE First International Conference on

Mobile Services, pages 33–39. IEEE.

Stadler, C., Lehmann, J., H

offner, K., and Auer, S. (2012).

Linkedgeodata: A core for a web of spatial open data.

Semantic Web, 3(4):333–354.

Stolz, A., Hepp, M., and Hemminger, A. (2017). Represen-

ting fashion product data with schema. org: Approach

and use cases. pages 254–272.

Uschold, M. and Gruninger, M. (1996). Ontologies: Princi-

ples, methods and applications. The knowledge engi-

neering review, 11(2):93–136.

Vansteenwegen, P., Souffriau, W., Berghe, G. V., and Oud-

heusden, D. V. (2011). The City Trip Planner: An

expert system for tourists. Expert Systems with Appli-

cations, 38(6):6540–6546.

Vansteenwegen, P. and Van Oudheusden, D. (2007). The

Mobile Tourist Guide: An OR Opportunity. OR In-

sight, 20(3):21–27.

GISTAM 2018 - 4th International Conference on Geographical Information Systems Theory, Applications and Management

246