THE GEOSPATIAL SEMANTIC WEB

Are GIS Catalogs prepared for This?

Carla Geovana N. Mac

ario

1,2

and Claudia Bauzer Medeiros

Institute of Computing, University of Campinas - UNICAMP, P.O.Box 6176, Campinas, SP, Brazil

Embrapa Agriculture Informatics, Embrapa, P.O.Box 6041, Campinas, SP, Brazil

Keywords:

Geospatial Data, Geospatial Semantic Web, GIS Catalogs, Interoperability on the Web.

Abstract:

Geospatial information catalogs are complex infrastructures that store and publish geographic information.

They are an important part of Geographic Information Systems (GIS), systems that manage geospatial data for

a wide variety of application domains. To be useful, a catalog must efﬁciently support discovery and retrieval

of geospatial information, working as a key component for planning and decision-making in a variety of do-

mains. Catalogs use standards to support data interoperability. However, the simple adoption of standards and

speciﬁcations for geospatial data description enables only syntactic interoperability. Semantic heterogeneity

still presents challenges for the so-called Geospatial Semantic Web. This work discusses some features that

GIS catalogs should have, focusing in semantic issues. We tested some existing and well known catalogs,

comparing them by means of these features. Based on this comparison, we identiﬁed some open issues that

should be addressed considering advanced Geospatial applications on the Web.

1 INTRODUCTION

The term geospatial data refers to all kinds of data

on objects and phenomena in the world that are asso-

ciated with spatial characteristics and that reference

some location on the Earth’s surface. Examples in-

clude information on climate, roads, or soil, but also

maps or telecommunication networks. Such data are a

basis for decision making in a wide range of domains,

ranging from studies on global warming to those on

urban planning or consumer services.

For example, geographic applications for con-

sumer services, like those provided by (Borges et al.,

2007) and (Jones et al., 2003), assign a location to

Web pages, based on existing geospatial evidences,

such as addresses and phone number. This informa-

tion can be subsequently used, for example, to ﬁnd

consumer services using fuzzy queries and to corre-

late Web pages spatially. In emergency management,

geospatial information can be useful to identify ar-

eas prone to disasters (Klien et al., 2004) or to help

in trafﬁc control. In agriculture they are very useful

for agroenvironmental planning (Mac

ario et al., 2007;

Mac

ario and Medeiros, 2008), providing means to en-

hance agricultural productivity.

The Web plays an important role in this sce-

nario, having become a huge repository of distributed

geospatial information. Data are collected and stored

by different organizations, which are required to ex-

change such data. These distributed data may be re-

trieved and combined in an ad hoc way, from any

source available in the world, extrapolating their local

context. Usually, the search for these data and meth-

ods is done by their syntactic content, focusing pri-

marily in keyword matching. This can lead to retrieval

of irrelevant data, and to omission of relevant facts.

Hence, semantic interoperability is also a key issue in

discovery, access and effective search for data in dif-

ferent application contexts. Solutions must take into

account the constant modiﬁcations in the real world,

and the evolution of our knowledge about the world.

There is a large amount of research on the

management of geospatial data, including proposals

of models, data structures, exchange standards and

querying mechanisms. One area of activity concerns

the so-called Geographic Information System (GIS)

catalogs. These work as metadata catalogs that can

be indexed by various means, such as by geographic

location, and provide support for users to search for

the data in different GIS data repositories. Catalogs

are based on a common set of ideas which do not take

semantic interoperability into account. This is a criti-

cal function necessary for advanced GIS applications,

specially in the context of the Geospatial Semantic

335

MacÃ ˛ario C. and Medeiros C.

THE GEOSPATIAL SEMANTIC WEB - Are GIS Catalogs prepared for This?.

DOI: 10.5220/0001839403350340

In Proceedings of the Fifth International Conference on Web Information Systems and Technologies (WEBIST 2009), page

ISBN: 978-989-8111-81-4

Web (Egenhofer, 2002). In this work we identify im-

portant criteria that must be met by catalogs. Based

on the results of comparing six widely used catalogs,

we point out issues for research and development in

the Semantic Web context. This discussion points at

directions that must be followed in order to enhance

the interoperability of GIS on the Web.

2 RELATED CONCEPTS

2.1 Geospatial Semantic Web

The Semantic Web was initially proposed by Berners-

Lee (Berners-Lee et al., 2001) as a way to bring struc-

ture to the meaningful content of Web pages, creating

an environment where users can obtain information

based on semantics and not only in syntax. In this sce-

nario, the Semantic Web would enable machines to

comprehend semantic documents and data, through:

(1) adoption of standardized data element names to

describe and exchange the data; (2) description of in-

formation in terms that allow common understanding;

(3) exposing data to be found and retrieved; (4) de-

signing efﬁcient retrieval mechanisms.

A standard establishes the name of data elements

(metadata) and/or groups of these elements, providing

a common set of terminology and deﬁnitions for the

description and exchange of data. The adoption of a

common vocabulary in this description ensures that

data producer and consumer share the same under-

standing of data. Hence, in the Semantic Web, the de-

scription of the meaning of data using ontology terms,

through standardized metadata is a way to provide se-

mantics, increasing interoperability. This description

process is called annotation.

The Semantic Web for geographic information,

called Geospatial Semantic Web by Egenhofer (Egen-

hofer, 2002), is a way to process requests involving

different kinds of geospatial information. This re-

quires the development of multiple spatial and do-

main ontologies, their representation in a way that

computers can understand and process, the process-

ing of queries considering these ontologies and the

evaluation of results based on the required semantics.

All of this leads to the search for a geospatial infor-

mation retrieval framework that relies on ontologies,

allowing users to retrieve desired data based on their

semantics.

In spite of several efforts, the Semantic Web is

far from becoming a reality (Shadbolt et al., 2006).

Although several standards have been developed and

adopted, there are too many variables that need to be

considered. The variety of user proﬁles and needs,

and of application domains – and thus of ontologies –

are just some of these factors. So far, most retrieval

engines are restricted to text, and other kinds of me-

dia pose countless challenges to the effective implan-

tation of the Semantic Web (Mac

ario and Medeiros,

2008).

2.2 Geospatial Catalogs

Catalogs are complex structures that enable data to

be found and retrieved, through the publishing of de-

scriptions of these data by metadata, known as anno-

tations (Nogueras-Iso et al., 2005), and operations on

these annotations. Catalogs offer search mechanisms

that access them to retrieve the desired data.

A GIS catalog is a Web application to publish de-

scriptions of geospatial data, enabling users to search

for the desired data (OCG, 2006). Because of stan-

dardized interface speciﬁcations, different users can

access them from all kinds of sites to search for the

content they need.

The Open Geospatial Consortium, OGC (OCG,

2006) is a non-proﬁt international organization that

is leading the development of standards for geospa-

tial and location based services. The consortium aims

at interoperability among geospatial systems, mak-

ing complex spatial information and services acces-

sible and useful to all kinds of applications. It de-

scribes three basic operations that a geographic cat-

alog should provide: publication, discovery and re-

trieval of geospatial metadata.

Geospatial data is described by metadata and these

descriptions are published in a catalog to support data

discovery. Data discovery can be performed either

by browsing the content of the catalog or by choos-

ing certain query terms. Once the desired metadata is

found, the data referenced can be retrieved.

3 DESIRABLE GIS CATALOG

FEATURES

In a Web environment, GIS users need to explore

available databases to discover the desired informa-

tion. In order to ﬁnd the data, the ﬁrst step is to search

for speciﬁc GIS catalogs and, once connected to the

catalog, look for candidate metadata describing the

desirable data. As the needed data is found, the users

can download and use it in theirs applications.

However, this is not an easy task to perform.

Geospatial data are complex, due to their spatial com-

ponent and its dynamic characteristics. Besides this,

users are hampered in their queries because of the

many different concepts and terms used to describe

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

336

data items. Catalogs seldom publish semantic anno-

tations. One possible approach for this is the use of

terms of an ontology to describe data, helping to re-

move the ambiguity. The increase in quality of the re-

trieved information and enhanced interoperability are

some beneﬁts from the adoption of semantic descrip-

tions, also known as semantic annotations. Although

there is extensive research in geospatial semantics,

it is focused mainly in the adoption of standardized

data element names and of ontology terms to describe

the data. It is not common to ﬁnd semantic catalogs,

which are those that publish semantic annotations and

support search on them as a way to enhance the re-

trieval of information. In this section we describe

the main features that a catalog should provide in or-

der to make the Geospatial Semantic Web a reality.

These features are based on those presented by (Lar-

son et al., 2006) and (ESRI, 2003), always consider-

ing the user viewpoint.

Feature 1: OGC Compliance. One of the many

standards proposed by OGC is the Catalog Services

Interface Standard (CAT), which supports the ability

to efﬁciently publish and search collections of meta-

data about geospatial data, services and related re-

sources. Hence, focusing in interoperability, a catalog

should be OGC compliant, enabling its use by users

and also by other catalogs.

Feature 2: Standards for Metadata. Catalogs

should support metadata standards. The growing need

for geospatial information led to the development of

a number of initiatives to obtain spatial metadata ac-

cording to a variety of formats within agencies, com-

munities of practice, or groups of countries. This re-

sulted in well established and widely used standards

like the ISO 19115 Metadata Standard (ISO, 2008),

or the FGDC geospatial metadata standard (FGDC,

1998). The objective of these standards is to provide

a common set of terms and deﬁnitions for the docu-

mentation and exchange of geospatial data.

The ISO 19115 standard (ISO, 2008) is a well

known standard for geographic information metadata

that deﬁnes the schema required for describing geo-

graphic information and services. It provides infor-

mation about the identiﬁcation, the extent, the qual-

ity, the spatial and temporal schema, spatial reference,

and distribution of digital geographic data (Silva,

2008). The Federal Geographic Data Committee

(FGDC, 1998) develops geospatial data standards for

implementing the USA National Spatial Data Infras-

tructure. The Content Standard for Digital Geospatial

Metadata (CSDGM), which is often referred to as the

FGDC Metadata Standard, provides the deﬁnition of

proﬁles and extensibility through user deﬁned meta-

data extensions.

Feature 3: Support Advanced Search. Catalogs

should provide different means for users to perform

their queries, considering different access levels to

each catalog and its contents. Users may perform the

search considering speciﬁc metadata elements, in a

way to reﬁne their query. It is a good choice to pro-

vide exploration tools, enabling users to explore the

retrieved data to determine suitability to their appli-

cations. Users should be able to select the desired

sources and categories and kinds of data to be re-

trieved. Besides this, it is important that each search

option be described, enabling its use by foreign peo-

ple. In this sense, the adoption of standard interfaces

can be very useful. Catalogs should also allow users

to view metadata records to determine if the retrieved

data is suitable for the intended use.

Feature 4: Save Data Online. Catalogs should al-

low users to view entire metadata records to deter-

mine if the corresponding data is suitable for the in-

tended use. Once the user ﬁnds the desired content

in a catalog, it is important to have means to save its

description or even the content itself. Hence, catalogs

should support a range of methods for online data de-

livery (e.g., live data streaming, commonly used data

formats, FTP download, and CDROM).

Feature 5: Provide Access to Multiple Servers. A

catalog should support search considering other meta-

data servers, increasing the number of repositories to

be searched. It has to be done in a consistent way,

enabling users to discovery new information reposito-

ries. The study (ESRI, 2003), shows that most users

do not perform distributed search due to problems on

catalogs. Instead, they go to speciﬁc GIS catalogs and

browse them to ﬁnd relevant data for their projects.

The portal should also support a search against a sin-

gle catalog.

Feature 6: Cater to Geospatial Data Diversity.

Geospatial data users are always looking for different

kinds of data, and also Web services. Hence, catalogs

should provide description of all these kinds of data,

allowing access to them. For example, maps should

be viewable in the browser or through an appropriate

software.

Feature 7: Support Semantic Search. Traditional

search mechanisms based on keyword matching are

restrictive. More expressive search algorithms, which

enhance recall and precision, should be available –

e.g., via thesauri, gazetteers and multilingual process-

ing. A more ﬂexible option is the use of ontology

terms to describe the data. In this sense, the catalog

should enable automatic matching of these terms dur-

ing the discovery process.

THE GEOSPATIAL SEMANTIC WEB - Are GIS Catalogs prepared for This?

337

4 COMPARING GIS CATALOGS

4.1 Overview of Selected Catalogs

We tested some GIS catalogs, as a means to identify

issues for research and development in the Semantic

Web context in order to enhance the interoperabil-

ity of GIS on the Web. Although these catalogs are

standardized interface speciﬁcations, they are imple-

mented considering different requirements, even for

the geographic domain. In this test we considered the

guidelines we stated in section 3.

Embrapa Information Agency (Souza et al.,

2006) is a Brazilian Web system to organize, deal

with, store, publish and access the technological in-

formation generated by Brazilian Agricultural Re-

search Corporation - Embrapa and other agricultural

research institutes. Knowledge is organized hierar-

chically, under the form of a tree. Although directed

to agricultural domain, knowledge is described using

Dublin Core metadata (Weibel et al., 1995), to allow

its retrieval by different user proﬁles. Only a syntactic

search for discovery of the stored resources is avail-

able, and search results can be saved in a textual ﬁle.

INSPIRE (www.inspire-geoportal.eu) is an Euro-

pean initiative that aims to provide geospatial infor-

mation to be used to formulate, perform and evaluate

european policies. Its objective is to create a spatial

information infrastructure to deliver integrated spatial

information services. The main users of INSPIRE in-

clude policy-makers, planners and managers at Euro-

pean, national and local level as well as the citizens

and their organisations.

FAO – The UNO Food and Agriculture Organiza-

tion leads international efforts to defeat hunger (FAO,

2008). The FAO catalog aims to share geographically

referenced thematic information between different or-

ganizations. It was implemented using the GeoNet-

work opensource (geonetwork-opensource.org), a

standards based, free and open source catalog applica-

tion to manage spatially referenced resources through

the web. It offers metadata editing and search func-

tions, as well as an embedded interactive web map

viewer. The catalog provides access to interactive

maps, satellite imagery and related spatial databases

maintained by FAO and its partners.

IDEE – Spatial Data Infrastructure of Spain

(www.idee.es) aims to integrate all data, metadata,

services and geographic information produced in

Spain. Its goal is to make the location, identiﬁcation,

selection and access of these contents an easier opera-

tion to their potential users. The IDEE catalog enables

users to search for geographic information – maps, or-

tophotos, etc – available for an area or a theme, in a

speciﬁc period of time.

GeoSpatial One Stop - GOS (gos2.geodata.gov)

is a public GIS catalog that aims to improve the ac-

cess to geospatial information and data. The catalog

is constructed under the U.S. Geospatial One-Stop E-

Government initiative for enhancing government efﬁ-

ciency and improving citizen services. Through the

catalog it is possible to ﬁnd data or map services,

make a map, browse community information, coop-

erate on data acquisition. Information is provided by

government agencies, individuals, and companies, or

obtained by harvesting the data from geospatial clear-

inghouses.

4.2 Comparison of Catalogs

Table 1 shows a comparative analysis of the presented

catalog systems, taking into account the features pre-

sented on section 3.

Except for the Embrapa Agency and GOS, all

the analyzed tools were implemented considering the

speciﬁcations provided by OCG. Though GOS is not

compliant with OGC, it was implemented according

to the National Spatial Data Infrastructure provided

by FGDC, which also focus on cooperative produc-

tion and sharing of geographic data. All the catalogs

provide data that are described using metadata stan-

dards, most of them using FGDC or ISO 19115. This

indicates that they all aim to promote the exchange

of the data they provide. However, to really support

data exchange, it is necessary that these descriptions

be supplied in an exchangeable format, like XML or

csv. The translation of element names from a stan-

dard, or saving data descriptions in a textual format,

as Embrapa Agency and IDEE do, restricts this ex-

changing.

The search for data is provided both in simple and

in advanced ways in all tested catalogs, except on

IDEE, which offers only the advanced one. A sim-

ple search enables the user to look for the keyword

occurrence within the entire record. However, this

can be a hard operation. Embrapa Agency, though

offering both kinds of search, has a limited number

of options for the advanced search. The same oc-

curs with IDEE. Only three of the catalogs provide

access to multiple GIS catalogs, supporting search in

different repositories. Though IDEE has this feature,

at present it accesses only the National Geographic

Institute data. All catalogs provide digital and non

digital data, but INSPIRE also provides search for ser-

vices and applications, which can improve the inter-

operability among geographic systems. Finally, none

of the systems analyzed enables a search based on the

semantics of the data.

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

338

Table 1: Evaluated GIS Catalogs.

5 OPEN RESEARCH TOPICS

This section summarizes some open research issues

that we have identiﬁed as a result of the comparison

presented on subsection 4.2. This reﬂects what we ex-

pect to be the most important features to be supported

by catalogs, towards making the Geospatial Semantic

Web a reality.

• Search on Multiple Servers. We identify this as a

challenge because of the following: (1) some cat-

alogs presented bad performance, thus motivating

the need to develop or adopt better algorithms; (2)

some results were very difﬁcult to interpret be-

cause of the language they use, making the data

useless. Hence, content description has to be also

in a well-known language; (3) some results were

dependent on available services. As many catalog

or data providers were ofﬂine, it was impossible

to get the data.

• Semantic Search. This is a central issue to be con-

sidered. The available catalogs do not provide this

kind of search, in spite of its usefulness when it

comes to geospatial data. A good survey of se-

mantic search approaches can be seen in (Man-

gold, 2007).

• Query Modiﬁcation. Although this is part of the

previous item, it is also an important issue to be

considered by itself. Query modiﬁcation in cata-

log search can help disambiguate search expres-

sions and enhance semantics.

• Adoption of Standards. This is a large ongoing

effort, focusing on interoperability of geospatial

data. The FAO Catalog and GOS are good exam-

ples for this issue. However, each one is based

on a different, but well known, geospatial stan-

dard. Hence, if their contents are to be combined,

one must develop translators from one to the other.

Common standards would avoid this kind of prob-

lem.

• Standard Interfaces. Once a user wants to search

for data in different catalogs, she has to identify

the available search options and what each ﬁeld

means. We identify the design of common in-

terfaces as a promising research area. The devel-

opment of standardized services can also enhance

the use of the available catalogs.

6 CONCLUSIONS

Geospatial data available on the Web are very useful

to answer important questions for various domains,

such as emergency management, services and agroen-

vironmental planning. Geographic catalogs are orga-

nized as descriptive lists of metadata, which describe

existing geospatial data. Through the publishing of

these metadata, users are allowed to search for the de-

sired information to be used in their systems. How-

ever, this search is not a trivial task, subject to a wide

range of problems. In particular, in the context of the

Geospatial Semantic Web, there are two main issues

to be addressed: (1) how to perform semantic search,

seen as a means to reduce the ambiguity of terms? (2)

what should be done in order to have a huge semantic

geospatial data network?

This work discussed features that GIS catalogs

should present, focusing in the Geospatial Semantic

Web. These features are based on interoperability is-

sues, from the user viewpoint. We tested some exist-

ing and well known GIS catalogs, comparing them by

means of these criteria. Furthermore, we identiﬁed re-

search and development issues that are not addressed

by the tested catalogs, and that are very important for

THE GEOSPATIAL SEMANTIC WEB - Are GIS Catalogs prepared for This?

339

advanced Geospatial applications. Although many of

the existing catalogs are good, they are far from what

is needed to support Semantic Networks. Much effort

has to be directed to the use of ontologies on search

operations. Distributed search also represents a chal-

lenge, as this is not a controlled operation. Finally,

the adoption of standard interfaces could facilitate the

search for data. Initiatives such as OGC are doing a

good work in this direction. However there are still

gaps to be ﬁlled.

ACKNOWLEDGEMENTS

We would like to thank Embrapa, FAPESP, Virtual

Institute FAPESP-Microsoft (eFarms project), CNPq

and CAPES for the ﬁnancial support for this work.

REFERENCES

Berners-Lee, T., Hendler, J., and Lassila, O. (2001). The

Semantic Web. Scientiﬁc American, pages 34–43.

Borges, K. A. V., Laender, A. H. F., Medeiros, C. B., and

C. A. Davis, J. (2007). Discovering geographic loca-

tions in web pages using urban addresses. In GIR ’07:

Proceedings of the 4th ACM workshop on Geograph-

ical information retrieval, pages 31–36, USA. ACM.

Egenhofer, M. J. (2002). Toward the semantic geospatial

web. In Proc. of the ACM GIS02, pages 1–4.

ESRI (2003). Implementing a metadata catalog portal in a

GIS network. Technical report, ESRI.

FAO (2008). FAO - GeoNetwork.

http://www.fao.org/geonetwork/srv/en/main.home.

FGDC (1998). FGDC-STD-001-1998. Content Standard

for Digital Geospatial Metadata. Washington, D.C.

ISO (2008). ISO 19115:2003 Geographic in-

formation – Metadata. ISO. Available

on:<http://www.iso.org/iso/home.htm>.

Jones, C. B., Abdelmoty, A. I., and Fu, G. (2003). Maintain-

ing ontologies for geographical information retrieval

on the web. In OTM Confederated International Con-

ferences - CoopIS, DOA, and OOBASE, pages 934–

951.

Klien, E., Einspanier, U., Lutz, M., and Hbner, S. (2004).

An architecture for ontology-based discovery and re-

trieval of geographic information. In 7th AGILE Con-

ference on Geographic Information Science, Herak-

lion, Greece. Parallel Session 3.1– Semantics I.

Larson, J., Siliceo, M., Silva, M., Klien, E., and Schade,

S. (2006). Are geospatial catalogues reaching their

goals? In 9th AGILE Conference on Geographic In-

formation Science - Poster, Visegrd, Hungria.

Mac

ario, C. G. N. and Medeiros, C. B. (2008). A framework

for semantic annotation of geospatial data for agricul-

ture. Int. J. Metadata, Semantics and Ontology - Spe-

cial Issue on ”Agricultural Metadata and Semantics”.

Accepted for publication.

Mac

ario, C. G. N., Medeiros, C. B., and Senra, R. D. A.

(2007). The webmaps project: challenges and results

(in portuguese). In IX Brazilian Symposium on GeoIn-

formatics - Geoinfo 2007, Brazil.

Mangold, C. (2007). A survey and classiﬁcation of semantic

search approaches. Int. J. Metadata, Semantics and

Ontology, 2:23–34.

Nogueras-Iso, J., Zarazaga-Soria, F., Bjar, R., lvarez, P. .,

and Muro-Med, P. (2005). OGC catalog services: a

key element for the development of spatial data in-

frastructure. Computers & Geosciences, 31:199–209.

OCG (2006). CSW 2.0 FGDC application proﬁle. Technical

Report OGC 06-129r1.

Shadbolt, N., Berners-Lee, T., and Hall, W. (2006). The

semantic web revisited. IEEE Intelligent Systems,

21(3):96–101.

Silva, H. (2008). MIG - Metadata for Geographic Informa-

tion –Introduction to ISO 19115 standard. Portuguese

Geographic Institute, Portugal. (in portuguese).

Souza, M. I. F., Santos, A. D., Moura, M. F., and Alves,

M. D. R. (2006). Embrapa information agency: an

application for information organizing and knowledge

management. In II Digital Libraries Workshop, pages

51–56, Brazil. (in portuguese).

Weibel, S., Godby, J., Miller, E., and Daniel, R.

(1995). OCLC/NCSA Metadata Workshop Re-

port. Web site http://www.oclc.org:5046/oclc/re-

search/conferences/metadata/

dublin

core report.html.

WEBIST 2009 - 5th International Conference on Web Information Systems and Technologies

340