A Model For Semantic Annotation Of Environmental Resources The Tatoo Semantic Framework

A Model for Semantic Annotation of Environmental
Resources: The TaToo Semantic Framework
Tomás Pariente, José María Fuentes, María Angeles Sanguino, Sinan Yurtsever,
Giuseppe Avellino, Andrea E. Rizzoli, Saša Nešić
ATOS Origin, Madrid, Spain
{tomas.parientelobo, jose.fuentesl, maria.sanguino,
sinan.yurtsever}@atosresearch.eu
Telespazio, Rome, Italy
giuseppe.avellino@telespazio.com
IDSIA, Lugano, Switzerland
{andrea, sasa}@idsia.ch
Abstract. In the last years we have seen the release of huge amount of
resources in the environmental domain. The use of non-formal tags for
annotating resources allows simple categorization and search capabilities, but it
does not provide the means to create cross-domain annotation and discovery.
On the other hand, ontologies are a shared and formal conceptualization of a
given domain. The use of formal semantics allows taking advantage of the
reasoning and inference power of the ontologies to create richer resource
annotations enhancing the discovery process. In the environmental domain
there is a clear need of frameworks and tools allowing formal tagging and
discovery. In this paper we discuss about the definition of a Semantic
Framework helping on the tagging and discovery process of environmental
resources. Moreover, we also report on the definition of a model to describe
environmental resources allowing cross-domain annotation and search.
Keywords: environment, discovery, annotation, ontology, cross-domain, search,
tagging.
1 Introduction
Since its inception, the Web has brought a revolution on how to publish, transmit and
consume information. While at first the number of information publishers on the Web
was limited, now almost everyone publishes personal and / or professional
information on the Web. Much of this growth is due to the fact that the expertise
needed to publish information on the Web is quite small. Basically, publishers only
need to own information and organize it in a way that makes it possible to be accessed
by consumers. However, the ease of publication, which originally facilitated the
growth of the Web, is not necessarily matching the ease of discovery and access to the
data. In professional environments, where information needs are highly complex, it is
necessary that machines, and not just humans, understand the information published

2 Tomás Pariente, José María Fuentes, María Angeles Sanguino, Sinan Yurtsever,
to improve its exploitation capabilities. To be able to allow for automated machine-
processing of information we need to introduce the concepts of meta-information and
semantics associated to it. And this is also the context for the TaToo project.
The TaToo project, which started in 2010, aims at improving the discovery and
exploitation capabilities of the published environmental information by providing a
way to semantically enrich environmental resources. TaToo is built on top of three
fundamental pillars:
A framework to create complex and formal annotations. Informal annotations are
relatively good for human interpretation, but fail to help machines to understand
the meaning and perform more advanced tasks based on those annotations.
Annotations should be then formal and shared in the domain.
Usage of ontologies to enhance discovery capabilities and to prevent
interoperability issues. TaToo provides an ontology framework that allows cross-
domain and cross-language tagging and search. Producing annotations is a time
consuming task. TaToo offers both the automatic harvesting of meta-information
from existing resources, and a manual interface for humans; both approaches are
based on the above mentioned ontology framework.
A community driven approach. TaToo pays special care in providing easy-to-use
and effective tagging interfaces to enable and facilitate the tagging process. TaToo
is strongly inspired by existing social bookmarking initiatives, such as reddit,
StumbleUpon, Digg, etc. In this sense, TaToo allows the user producing their own
annotations, and share them with the community, starting an information
enrichment cycle. On the other hand, TaToo differs from the aforementioned social
approaches as it is using formal semantic annotations.
This paper focuses on the TaToo ontology framework. The first objective of TaToo is
the establishment of an ontology framework that allows the production of formal
resource descriptions. Subsequently, the TaToo system is taking advantage of domain
ontologies and formal resource descriptions to improve the exploitation of
environmental resources by improving discovery processes, enhancing the resource
presentation and providing interoperability between different fields of the
environmental domain with the aim of allowing cross-domain search.
Much of the functionality provided by TaToo relies on ontologies. Ontology
development is a complex, time-consuming task involving different roles (ontology
developers and domain experts) and activities (requirements acquisition, design, etc.).
Therefore, from the ontological point of view, one of the biggest challenges of TaToo
is to use or produce a formal description of the environmental domain that is at the
same time good enough to cover the project objectives, and on the other hand simple
enough to avoid unnecessary complexity and gruesome maintenance. The
environmental domain is actually a very broad domain that includes several sub-
domains dealing with such diverse areas as climatology, soil composition, crop
management, etc.
Although there are ontologies dealing with environmental issues, presently there is
no encompassing ontology detailing such a broad domain that fits the needs of TaToo.
Therefore, a realistic objective of TaToo in this regard is to allow sub-domain experts

A Model for Semantic Annotation of Environmental Resources: The TaToo Semantic
Framework 3
to model their own sub-domain and at the same time provide the maximum degree of
integration between such sub-domain models. Because of these reasons the ontology
framework should ensure semantic interoperability between different domain
ontologies while ensuring their modular nature.
2 Ontology Framework
There are several approaches to achieve semantic interoperability when dealing with
different ontologies. Watche [2] defined three ways to integrate ontologies by using
single ontology, multiple ontology or hybrid ontology approaches.
The single ontology approach uses one global ontology that provides a shared
vocabulary for the specification of the semantics. Thus, all information sources are
related to one ontology, which usually is a combination of several specialized
ontologies. This approach is suitable for solving integration problems where all
information sources to be integrated provide nearly the same view on a domain.
In the multiple ontology approach each information source is described by its own
ontology. The main advantage of this approach is that no common and minimal
ontology commitment about a global ontology is needed. To achieve a common
understanding of the information sources, defined ontologies must be related. To
relate several “source describing” ontologies, several inter-ontology mappings must
be introduced. In practice, these inter-ontology mappings are very difficult to define
because of the many semantic heterogeneity problems which may occur, and the
situation is aggravated when there are many domain ontologies to map.
The hybrid approach is based on using a common shared ontology and a set of
local or application ontologies that are mapped uniquely to the shared vocabulary. In
this case the local ontologies would extend the vocabulary to the needs of a given
domain and the interoperability is achieved based on the shared ontology. TaToo
follows the hybrid approach as shown in Fig. 1.
Fig. 1.TaToo high-level ontology framework

In order to apply the hybrid approach, a bridge ontology that will become the upper
layer of the integrated ontology framework has to be developed. Our bridge ontology
includes concept and property definitions from widely adopted ontologies and
necessary concept definitions in order to ensure the mappings between domain
ontologies. Any domain ontology should be mapped properly to the bridge ontology
in order to be usable by the integrated ontology framework. So the system interacts
with the domain ontologies via the bridge ontology.
The mapping process is done manually by defining equivalency and subsumption
relationships between concepts, object properties, data properties and individuals of
bridge ontology and the domain ontology. In the end from the point of view of the
bridge ontology, the entire ontology framework is seen as one big knowledge base. In
Fig. 2, where the current version of the TaToo ontology framework is shown, some
example mappings between ontology elements from the bridge ontology (x-shaped
circles in green) and the domain ontologies (x-shaped circles is gray) are depicted to
represent the cross-domain alignment process.
Fig. 2. Current version of the TaToo ontology framework
Some of the selected ontologies included in the TaToo bridge ontology are
GeoNames, an ontology for representing geospatial expressions, and OWLTime, an
ontology for representing temporal expressions. There are already implemented
reasoners which especially support both ontologies for geospatial and temporal
reasoning. One of them is OWLIM, an OWL semantic repository, which supports
geospatial reasoning exclusively for the GeoNames ontology. As TaToo is expected
to offer users the possibility to create time and location based annotations and
temporal and spatial queries, these ontologies become an essential part of the bridge
ontology. In particular, OWLIM has been selected to be used as the semantic
repository and the availability geospatial reasoning has been one of the reasons for
this choice. OWLIM supports OWL2-RL, a web ontology language profile inspired

Framework 5
by rule based languages. OWL2-RL is much more efficient and scalable compared to
OWL2-DL but less expressive. This is the reason why OWL2-RL has been chosen.
Because of the constant incremental annotated data that will be manipulated, a need
for more efficient reasoning arose and OWL2-RL is aimed to cover these kinds of
needs.
In TaToo, there is also the need to uniformly describe environmental information
resources. We think of an environmental resource as a web resource (being a web
page, a document, a model, a service, etc.), which is identified by a URI. In this sense,
the Minimal Environmental Resource Model (MERM) is defined as a part of the
shared ontology that will contain the minimal set of cross-domain concepts and
properties. For future releases, the usage of an upper-level ontology to perform more
elaborate cross-domain mappings will be investigated. The Semantic Web for Earth
and Environmental Terminology (SWEET) [1], a widely-used set of ontologies
providing semantic descriptions of Earth system science, is a good candidate for the
TaToo ontology framework.
3 The Minimal Environmental Resource Model
A minimum resource model is defined as the largest common denominator between a
set of heterogeneous description formalisms related to a common resource. Minimum
resource models have been applied in other research areas with satisfactory results. In
[4] a description of an open system for web services publication and discovery based
on a description of a minimal service model resulting in an ontology for service
modelling called POSM is presented. Functionally, the minimum resource model
aims at a similar objective to the bridge ontology: providing a common framework
between different domain ontologies. However, conceptually, the minimum resource
model is an effort to identify a minimal model that, without limiting the expression of
specific domains, acts as a reference for past and future applications in a specific area.
This effort and referencing aim is what differentiates the bridge ontology of a
minimum resource model and is what makes it a valuable resource.
One of the most relevant parts of the TaToo bridge ontology is the Minimal
Environmental Resource Model (MERM). MERM describes the structure of
resources and annotations in TaToo, being a reference data model for both services
and user interfaces. As any other ontology developed in the scope of TaToo, MERM
has been developed following the NeOn methodology [3]. The NeOn methodology
encourages reusing as much as possible. To that extent, MERM contains, besides the
already mentioned POSM ontology, concepts from several other ontologies like
SIOC1
, FOAF2
, Dublin Core3
and O&M4
. Fig. 3 presents the main classes of MERM:
Resource, Annotation, and ResourceAccessInfo.
1
The SIOC initiative (Semantically-Interlinked Online Communities), http://www.sioc-
project.org/
2
The Friend of a Friend (FOAF) project, http://www.foaf-project.org/
3
Dublin Core metadata iniciative, http://dublincore.org/

Fig. 3. MERM core classes and relations
The Resource class represents a resource in the TaToo system. It includes resource
management information as author, owner, provider, date of creation, etc. A Resource
usually has access information presented in the ResourceAccessInfo class. The
ResourceAccessInfo class contains all information needed to access a resource. This
information is very heterogeneous ranging from a simple URL to a complex WSDL
depending on the nature of the resource. Finally, a resource presents a set of
annotations, represented by the Annotation class. An annotation describes what a
resource is about. Except for some management information, the content of an
annotation depends on the type of the annotated resource. An annotation for a web
document would be different from an annotation for a web service, being the “some
resource is related to some domain concepts” the simplest kind of annotation.
Resources in TaToo are retrieved taking their annotations as basis. In the current
version, MERM distinguishes between three kinds of annotations:
TimeSeriesAnnotation that describes time series following the model proposed by
the Observations and Measurements Ontology.
WebServiceAnnotation that describes web services following the model proposed
by POSM.
4
Observations and Measurements ontology, http://seres.uni-
muenster.de/o&m/O&M_discussion_paper.pdf

Framework 7
WebAnnotation that describes web resources following the model proposed by
SIOC.
MERM is still work in progress so the addition of new models for several
resources is expected. Fig. 4 shows in details the model adopted for
TimeSeriesAnnotation.
Fig. 4.TimeSeriesAnnotation model
Finally, there is also a special kind of annotations, the evaluations, which are
annotations that relates to another annotation instead to a resource. Evaluations are
created by users in order to evaluate how accurate or useful a resource annotation is.
This information is used during the discovery process to improve discovery results
and accuracy.
4 Integrating Domains
The TaToo Ontology framework, as it was described in Section 2, comprises the
bridge ontology, a number of domain ontologies, and a number of alignment
ontologies (see Fig. 2). The framework is not limited by a number of different domain
ontologies, that is, by a number of different environmental sub-domains whose
resources are managed by the TaToo system. However, in order to be able to plug-in a
domain ontology in the TaToo ontology framework, the domain ontology needs to be

accompanied by an appropriate mapping interface. This mapping interface is
identified in the framework as an alignment ontology. In this section we describe two
possible alignment/mapping strategies on the example of three domains ontologies,
specific to the validation scenarios of the TaToo project, namely the JRC, MU, and
AIT ontologies. The JRC ontology describes the agro-environmental domain, the MU
ontology describes the anthropogenic impacts of global climate change, and finally
the AIT ontology contains concepts related to the comparison of climatic conditions
in different regions. These three domain ontologies are considered to be an integral
part of the TaToo ontology framework thus justifying the proposed ontology
integration approach. Moreover, we also plan to use them in the evaluation of the
semantic tagging and discovery capabilities of the TaToo system.
The first mapping strategy, adopted for the JRC ontology, uses the rdfs:subClassOf
construct defined in the RDFS to map concepts of the domain ontology to the
concepts of the MERM and bridge ontology. For example, concepts describing some
domain specific environmental resources should be specified as sub-concepts of the
merm:Annotation concept or its sub-concepts (merm:TimeSeriesAnnotation,
merm:WebService-Annotation, and merm:WebAnnotation). Moreover, the concepts
introduced to conceptualize some domain specific features (topics), which are
supposed to be a part of the resources‟ annotations, should be defined as sub-concepts
of the bridge:Topic concept. To illustrate this mapping strategy on a practical
example, in Figure 5 we list a snippet of the JRC ontology mapping specifying a
resource annotation type that describes software development kits:
The second mapping strategy, used for the AIT and MU ontologies, assumes the
usage of the owl:equivalentClass and owl:sameAs constructs defined in OWL, plus
the skos:broadMatch and skos:narrowMatch constructs defined in the SKOS5
(Simple
Knowledge Organization System) vocabulary to map concepts and properties of a
domain ontology to the concepts and properties of the MERM and bridge ontology.
The difference between the OWL and the SKOS constructs is that the latter ones
provide a looser mapping. There is also a slight difference between the two OWL
constructs regarding the strength of the mapping they provide. The meaning of the
owl:equivalentClass is that the two concepts have the same set of individuals.
However, the owl:equivalentClass does not imply the concept equality. Concept
equality means that the concepts have the same intentional meaning. Real concept
equality can be expresses by the owl:sameAs construct. In Figure 6 we show a snippet
of the MU ontology mapping that specifies the mapping of the “Diagnose” concept
from the MU domain to the merm:Annotation concept of the MERM ontology.
5
http://www.w3.org/TR/skos-reference/
Fig. 5. A snippet of the JRC ontology mapping

Framework 9
The development of both bridge ontology (including MERM) and domain
ontologies (i.e., JRC, AIT, and MU) in the TaToo Ontology framework is an
evolutionary process. In case of domain ontologies we plan to extend them in future
by adding new domain specific concepts and relationships that are currently not
provided. In case of the bridge ontology we want to identify which of the concepts
from the domain ontologies can be generalized and potentially become members of
the MERM ontology.
5 Conclusions
In this paper we have presented the approach of the TaToo project to the
development of an ontology framework for the semantic annotation of environmental
resources. As the environmental domain is composed by a wide number of sub-
domains, we chose a hybrid approach to be able to integrate ontologies for the various
sub-domains. Such an approach required the development of a bridge ontology and a
series of alignment ontologies to map concepts from domain specific ontologies to the
bridge ontology. A minimum environmental resource model has also been developed
to be included in the bridge ontology in order to provide a minimal set of common
concept to be shared across the different domains and facilitate alignments.
The TaToo ontology framework is functional to the declared objectives of the
TaToo project that is to allow for the semantic annotation of environmental resources
in order to enhance and facilitate search and discovery.
Acknowledgements
The research leading to these results has received funding from the European
Community's Seventh Framework Programme (FP7/2007-2013) under Grant
Agreement Number 247893.
References
1. Raskin, R.G., Pan, M.J., 2005. Knowledge representation in the semantic web for
earth and environmental terminology (SWEET). Computers & Geosciences 31, 1119-
1125.
2. Wache H., Vogele T., Visser U., Stuckenschmidt H., Schuster G., Neumann H.,
Hubner S., 2001. Ontology based integration of information: a survey of existing
approaches. In: Proceedings of the IJCAI„01:17th International Joint Conferences on
Artificial Intelligence, Seattle, WA, USA, pp.108–117.
3. Gómez-Pérez A., Motta E., Suárez-Figueroa M.C., NeOn Methodology in a Nutshell.
http://www.neon-project.org/nw/NeOn_Book, 2010.
Fig. 6. A snippet of the MU ontology mapping

4. Pedrinaci C., Liu D., Maleshkova M., Lambert D., Kopecky J., and Domingue J.:
iServe: a Linked Services Publishing Platform. Workshop: Ontology Repositories
and Editors for the Semantic Web at the 7th Extended Semantic Web Conference
(ESWC) 2010. Heraklion, 2010.

A Model For Semantic Annotation Of Environmental Resources The Tatoo Semantic Framework

More Related Content

Similar to A Model For Semantic Annotation Of Environmental Resources The Tatoo Semantic Framework (20)

More from Andrew Molina (20)

Recently uploaded (20)

A Model For Semantic Annotation Of Environmental Resources The Tatoo Semantic Framework