=Paper=
{{Paper
|id=None
|storemode=property
|title=Conceptual Modeling of Formal and Material Relations Applied to Ontologies
|pdfUrl=https://ceur-ws.org/Vol-1041/ontobras-2013_paper19.pdf
|volume=Vol-1041
|dblpUrl=https://dblp.org/rec/conf/ontobras/LinckSA13
}}
==Conceptual Modeling of Formal and Material Relations Applied to Ontologies==
https://ceur-ws.org/Vol-1041/ontobras-2013_paper19.pdf
Conceptual Modeling of Formal and Material Relations
Applied to Ontologies
Ricardo Ramos Linck, Guilherme Schievelbein and Mara Abel
Institute of Informatics – Universidade Federal do Rio Grande do Sul (UFRGS)
Caixa Postal 15.064 – 91501-970 – Porto Alegre – RS – Brazil
{rrlinck,gschievelbein,marabel}@inf.ufrgs.br
Abstract. Ontologies represent a shared conceptualization of a knowledge
community. They are built from the description of the meaning of concepts,
expressed through their attributes and their relationships. Relationships are
used to describe how the concepts are structured in the world. This work
reviews the literature on formal and material relations, especially on
mereological and partonomic relations, and proposes an alternative for the
conceptual modeling of such relations in a domain ontology. This alternative
has been made available in the ontology building tool of the Obaitá Project.
1. Introduction
This work falls in the area of Conceptual Modeling and Knowledge Engineering,
focusing on the ontological foundations and conceptual modeling of relations applied to
ontologies.
Ontology represents a shared conceptualization that includes concepts, its
attributes and the relationships between the concepts. In addition to the subsumption
relationships that build the taxonomies of concepts, other formal and material relations
assist in structuring the domain and the conceptual definition. The main existing
modeling tools, such as Protégé, WebODE and others, however, are still deficient in
differentiating the various types of formal and material relationships in order to assign
the possibilities of automated reasoning.
Obaitá Project is a tool for collaborative construction of visual domain
ontologies based on foundational ontology. Continuing the development of the Obaitá
ontology building tool, this work provides support to the ontological foundations of the
relations, enforcing ontological consistency and providing visual component support
into the ontology relations.
The main goals of this research project include providing:
- foundation ontological constructs to support the ontological choices of the
kinds of relations through the semantic expressiveness of a foundational ontology,
especially the formal (mereological and partonomic) and material relations;
- support to the inference of the ontological meta-type of the relations based on
the meta-types of the respective related concepts;
- visual ontological constructs to represent the visual knowledge about relations
among the ontology concepts, supporting imagistic domains;
- intuitive interface which, through the use of natural language, does not require
users to have any prior knowledge of ontological representation formal languages.
185
Following, in Section 2, we present an overview of ontology and relations; in
Section 3, we present an analysis on some of the main ontology building tools; in
Section 4, we present our implemented solution; in Section 5, we present an example of
use of the system; and finally, in Section 6, we present our conclusions about this work.
2. Ontology and Relations
In recent years, there has been a growing interest in the use of foundational ontologies
for evaluating conceptual modeling languages, developing guidelines for their use and
providing real-world semantics for their modeling constructs (Guizzardi and Wagner,
2010). One of the main foundational ontologies is UFO (Unified Foundational
Ontology), which is divided into three incrementally layered compliance sets: UFO-A
defines the core of UFO, as a comprehensive ontology of endurants; UFO-B defines, as
an increment to UFO-A, terms related to perdurants; UFO-C defines, as an increment to
UFO-A and UFO-B, terms related to the spheres of intentional and social entities
(Guizzardi et al., 2007).
The importance of conceptual relationships is highlighted by (Bala and Aghila,
2011) when they state that relationships are fundamental to express semantics in
ontology in order to associate concepts and associate instances. Relationships are
defined according to their properties, like reflexivity, symmetry, transitivity. As argued
in (Guarino, 2009) and (Grenon, 2003), relations can be divided into two broad
categories, namely formal and material relations. Formal relations hold between two or
more entities (relata) directly, without any further intervening individual. Figure 1
exemplifies a formal relation between alcohol and wine, where alcohol is part of wine.
Figure 1. An example of formal relation.
Four sorts of conceptual formal part-whole relations are defined in (Guizzardi,
2005) with different semantics, based on the type of the related entities: component-of
relates individuals that are functional complexes, subquantity-of relates individuals that
are quantities, subcollection-of relates individuals that are collectives, and member-of
relates individuals that are functional complexes or collectives (as part) and a collective
(as a whole).
Parthood relationships are especially important for modeling visual knowledge,
since the object recognition by cognitive systems that support vision is strongly based
on composition and decomposition operations.
Unlike formal relations, material relations have material structure of their own
and include examples such as working at; for a material relation of being treated in
between Paul and a medical unit to exist, another entity must exist which mediates Paul
and the medical unit. These entities are named relators (Guizzardi and Wagner, 2010).
Figure 2 depicts an example of material relation between employee and company
(relata), where, if an employee works for a company, another entity (relator), such as
employment, must exist in order to mediate them.
186
Figure 2. An example of material relation.
3. Relations in Ontology Building Tools
The main ontology building tools, such as Protégé and WebODE, among others, provide
different support for specifying the ontology relations.
3.1. Protégé
Protégé 4 (Horridge et al., 2007) can compute subsumption relationships between
classes, and detect inconsistent classes. It can be computed automatically by a reasoner.
Binary relations, linking two individuals together, are represented by slots.
Properties describe binary relationships. There are two main types of properties:
datatype properties and object properties. Datatype properties describe relationships
between individuals and data values, and object properties describe relationships
between individuals. Object properties may define some characteristics, such as
functional, inverse functional, transitive, symmetric, asymmetric, reflexive, irreflexive.
Properties may present some restrictions, which fall into three main categories:
quantifier, cardinality and hasValue restrictions. The quantifier restrictions effectively
put constraints on the relations that the individual participates in. It does this by either
specifying that at least one kind of relationship must exist (existential restrictions), or by
specifying the only kinds of relationships that can exist (universal restrictions). The
cardinality restrictions are the number of relationships that an individual may participate
in for a given property. Cardinality restrictions may specify the minimum and the
maximum cardinality restrictions. The hasValue restrictions describe the class of
individuals that have at least one relationship to another specific individual.
3.2. WebODE
WebODE (Arpírez et al., 2001) allows the post-processing of the ontology, using the
OntoClean methodology for identifying incorrect taxonomic (is-a) relations. WebODE
works with both built-in relations and ad-hoc relations.
Built-in relations are predefined relations related to the representation of
taxonomies of concepts and mereology relationships between concepts. They are
divided into three groups: taxonomical relations between concepts, taxonomical
relations between groups and concepts, and mereological relations between concepts.
The taxonomical relations between concepts have two predefined relations: subclass-of
and not-subclass-of. Single and multiple inheritance are allowed. The taxonomical
relations between groups and concepts have two predefined relations: disjoint-subclass-
partition and exhaustive-subclass-partition. The mereological relations between
187
concepts have two predefined relations: transitive-part-of and intransitive-part-of.
Ad-hoc relations are characterized by their name, the source and target concepts
name, and its cardinality. WebODE allows just binary ad-hoc relations to be created
between concepts. The creation of relations of higher arity must be made by reification.
3.3. Remarks about the tools
Analyzing the available tools, we noticed that most of them have both implementation
and user interface oriented to formal languages of representation, like OWL, making it
harder for users who do not have this expertise to use them properly. We also noticed
that these tools do not include ontological foundations or visual domains.
The analyzed tools do not provide adequate support to the ontological choice
problem: how to choose the best primitives to represent the needed relations. These
issues may produce different specifications for the same conceptual model, or result in
different interpretations of the same model by different users. Likewise, the construction
of the relations in the user mind is strongly based in visual knowledge, but this topic is
still incipient for the main ontology building tools. In the next section we describe the
solution that has been implemented in order to achieve the goals of this research project.
4. Implemented Solution
This work supports the relation ontological foundations (according to UFO-A), enforces
ontological consistency, provides inference, and provides visual component support.
Relations are specialized in formal or material relations, as seen in Figure 3.
Material relations contain a relator and two relata. Formal relations contain two relata.
The relata are existing concepts from the domain ontology, and the relator is a
relational moment. Formal relations may be further specialized as part-whole relations
(component-of, member-of, subcollection-of or subquantity-of), enforcing the following
constraints: component-of, both relata are functional complexes (kind), they have to be
irreflexive, asymmetric and nontransitive, and they have weak supplementation;
member-of, the whole individual is a collective, while the part can be either a collective
or a functional complex (kind), they have to be irreflexive, asymmetric and intransitive,
and they have weak supplementation; subcollection-of, both relata are collectives, they
have to be irreflexive, asymmetric and transitive, and they have weak supplementation;
and subquantity-of, both relata are quantities, they have to be irreflexive, asymmetric
and transitive, they have strong supplementation, and they have to be nonshareable.
When editing a concept relation, it is possible to choose its name, its type
(classification by UFO-A), the target concept, the source and target cardinalities, the
relator (for material relations) and its icon (visual component). The source concept is
automatically selected as the concept that is being viewed in detail in the system.
In order to help users to define the relation type, the system guides them by
asking questions, without requiring users to have any knowledge of ontological
representation. For example, if he/she answers the question telling the system that the
relation needs the existence of a mediating entity, then the relation type is “material”.
The system also has the ability to infer the relation type based on the meta-types
of the respective related concepts. For example, if the meta-type of both related
concepts is “quantity”, then the relation type is “subquantity-of”. Next, we present an
188
example of use through a real domain ontology, from the Sedimentary Stratigraphy area,
in order to evaluate our research project proposed approach.
Figure 3. The meta-ontology relation structure.
5. Example of Use
In order to validate the system in a real environment, we brought an example of use in
the Sedimentary Stratigraphy domain, an area of Geology responsible for studying the
formation processes of sedimentary rocks. In (Lorenzatti et al., 2010), a domain
ontology was built with the help of experts, serving as the basis for initiating the system.
This domain has been chosen because it presents some important aspects for our focus:
it is strongly based on visual knowledge; its structure is complex; and it has scientific
and economic relevance, studying the generation and depositional conditions of
important mineral deposits, such as coal and oil.
From this example of use, we intend to evaluate the approach proposed by our
research project, considering the following parameters:
- Total of existing relations: “before” and “after” the activities performed by the
geologists, classified by relation type (only “after” the activities performed by the
geologists; the previous ontology relations were not based on ontological foundation).
- Total of changes performed on the relations: added relations, updated relations
and removed relations, classified by relation type.
After these evaluations, then it will be possible to identify the contributions and
resulting benefits from this research project approach regarding the ontological
consistency of the created ontology concept relations. In the next section, we present
our conclusions and some open possibilities for future improvement of this work.
6. Conclusions
The main contributions of this work include the definition of the ontological relations
based on a set of metadata, providing specialized ontological constructs for creating the
domain ontology relations and supporting the inference of the relation ontological meta-
types. The ontology building environment is independent of the representation formal
languages, providing intuitive interface so that users do not need any previous
ontological representation knowledge in order to interact with the ontology. Some
constructs allow the association of icons in order to obtain a higher domain
understanding. This work takes in consideration the importance of the relation
ontological foundations and the visual knowledge as supporting instruments.
189
As a result of our researches, our ontology building tool is constantly under
improvement; we keep adding important features on its implementation, which many of
them we do not find on most of the other tools. Thus, this specific research project has
fundamental importance, continuing the evolution of an innovative tool for both
academic and commercial purposes. An extensive evaluation on the modeling of the
ontology relationships still has to be performed, as described in the previous section. Its
benefits have already become explicit through the conceptual and intuitive approach
added to the tool. The capabilities of the proposed metadata model will be assessed
through a practical application by the construction of an ontology for the Sedimentary
Stratigraphy domain from Geology.
This work can be considered as a step for future work in order to complement
the ontological foundation of relations into the Obaitá ontology building tool, such as
taking in consideration the taxonomic relations and the temporal relations.
Acknowledgement
This work is supported by the Brazilian Research Council (CNPq) and Petrobras PFRH.
References
Arpírez, J. C., Corcho, O., Fernández-López, M. and Gómez-Pérez, A. (2001).
“WebODE: A Scalable Workbench for Ontological Engineering”. In: Proceedings of
the First International Conference on Knowledge Capture (K-CAP).
Bala, P. S. and Aghila, G. (2011). “Relationship Based Reasoning”. In: International
Journal of Computer Applications.
Grenon, P. (2003). “The Formal Ontology of Spatio-Temporal Reality and its
Formalization”. In: AAAI Technical Report Series, AAAI Spring Symposium on the
Foundations and Applications of Spatio-Temporal Reasoning, Stanford University,
Palo Alto, California, U.S.A.
Guarino, N. (2009). “The Ontological Level: Revisiting 30 Years of Knowledge
Representation”. Conceptual Modeling: Foundations and Applications, LNCS 5600,
Springer-Verlag.
Guizzardi, G. (2005). “Ontological Foundations for Structural Conceptual Models”.
PhD Thesis, University of Twente, The Netherlands.
Guizzardi, G. and Wagner, G. (2010). “Using the Unified Foundational Ontology (UFO)
as a Foundation for General Conceptual Modeling Languages”. Theory and
Applications of Ontology: Computer Applications, p.175-196, Springer.
Guizzardi, R. S. S., Guizzardi, G., Perini, A. and Mylopoulos, J. (2007). “Towards an
Ontological Account of Agent-Oriented Goals”. Software Engineering for Multi-
Agent Systems V, Springer.
Horridge, M., Jupp, S., Moulton, G., Rector, A., Stevens, R. and Wroe, C. (2007). “A
Practical Guide to Building OWL Ontologies Using Protégé 4 and CO-ODE Tools”.
Edition 1.1, University of Manchester.
Lorenzatti, A., Abel, M., Fiorini, S. R., Bernardes, A. K. and Scherer, C. M. S. (2010).
“Ontological Primitives for Visual Knowledge”. In: A.C. da Rocha Costa, R.M.
Vicari, F. Tonidandel (Eds.): SBIA 2010, LNAI 6404, p.1-10, Springer-Verlag.
190