=Paper=
{{Paper
|id=None
|storemode=property
|title=Conceptual relations established by processes: the case of cocción (firing) in a corpus of industrial ceramics
|pdfUrl=https://ceur-ws.org/Vol-578/paper7.pdf
|volume=Vol-578
|dblpUrl=https://dblp.org/rec/conf/tia/GarciaC09
}}
==Conceptual relations established by processes: the case of cocción (firing) in a corpus of industrial ceramics==
https://ceur-ws.org/Vol-578/paper7.pdf
TIA’09
Conceptual relations established by processes:
the case of cocción (firing) in industrial ceramics
Nava Maroto García1
Amparo Alcina Caudet2
1
Centro de Estudios Superiores Felipe II, Universidad Complutense de Madrid,
mnmaroto@cesfelipesegundo.com
2
Universitat Jaume I, Castelló, TecnoLeTTra Team
alcina@trad.uji.es
http://tecnolettra.uji.es
Abstract: This paper presents an analysis of some of the conceptual relations
that can be established by processes in the field of ceramic industry. First, the
conceptual nature of industrial processes is analysed. Second, we define
argumental and circumstantial relations in which one of the elements linked by
the relation belongs to the conceptual class activity. Then, we present a case
study of the concept “cocción” (firing) in the context of ceramic industry and
finally, we put forward some concluding remarks about the relationships
established by processes.
Key-words: processes, conceptual relationships, ontologies, ceramic industry
Résumé: Ce papier présente une analyse de certaines relations conceptuelles
qui peuvent être établies par des processus dans le domaine de l’industrie
céramique. D’abord on analyse la nature conceptuelle des processus. En
deuxième lieu, nous présentons des relations argumentales et circonstancielles
dans lesquels un des éléments uni par le lien pourrait correspondre à la classe
conceptuelle activité. Ensuite, nous présentons un cas d’étude du concept «
cocción » (cuisson) dans le contexte de l’industrie de la céramique et
finalement nous avançons quelques conclusions sur les liens établis par les
processus.
Mots-clés: Processus, Relations conceptuelles, Ontologies, Industrie céramique
1 Introduction
In this contribution we study the nature of the conceptual relationships established
by concepts belonging to the class activity, in particular, by industrial processes. In
order to do so, we set out from the catalogue of conceptual relationships developed in
our previous research (Maroto, 2007; Alcina, 2009; Maroto & Alcina, 2009).
Conceptual relations established by processes
This research is part of the TXTCERAM1 and ONTODIC2 projects, which are being
carried out by the TecnoLeTTra Team at the Universitat Jaume I in Castellón (Spain).
The TXTCERAM project’s main objective is to create an electronic corpus of
specialized texts from the field of ceramics which can be used to test the efficiency of
certain software tools in the design of an integrated computer-assisted system for
elaborating and consulting terminologies. The aim of the ONTODIC project is to
propose a systematic methodology for the elaboration of onomasiological
terminological dictionaries using an ontology editor. As part of these projects we are
currently developing a dictionary of ceramics terminology that allows the user to
make queries based on the meaning and not only through the lemma.
In this context, the formal representation of conceptual relations plays a key role
when we want to retrieve information about concepts. For example, we may want to
retrieve all the floor tiles produced through a particular process (for example,
extrusion). If we have formally represented the relationship between each type of
floor tile and its production process through the conceptual relation product-process,
then we will be able to retrieve this information. In Maroto & Alcina (2009) we
applied our proposal for the formalization and retrieval of conceptual information
about relationships established by finished ceramic goods. In this contribution we
focus on the representation of conceptual relationships established by the process of
firing. Processes are complex concepts, and therefore we consider it worthwhile
analysing whether our proposal for the formal representation of finished ceramic
goods is also suitable in the case of industrial processes.
First of all, activity is defined as one of the conceptual classes in the model put
forward by Sager and Kageura (1994) and is related to the description of “dynamic
concepts” proposed by Pilke (2001) and taken up by Nuopponen (2007).
Second, we briefly describe the catalogue of conceptual relationships developed
in our previous research, with special emphasis on argumental and circumstantial
relationships, in which at least one of the concepts involved is a process, and
therefore belongs to the conceptual class activity.
Then we explain the methodology followed and the preliminary results obtained
in our empirical study of the concept “cocción” (firing). For this empirical study we
have used both corpus and manual analysis of specialized texts about ceramic
industry. We will point at some difficulties that arise when we represent knowledge
about the relationships established by processes.
Finally, our model for the implementation of “cocción” in Protégé is presented,
and we discuss the results and provide some concluding remarks about the specificity
of activities with a view to implementing them in an ontology editor.
1
“TXTCERAM. Extracción semiautomática y análisis conceptual formal de términos de la cerámica a partir
de un corpus electrónico. Su eficacia y utilidad en la mediación lingüística”. (TXTCERAM. Semiautomatic
extraction an formal conceptual analysis of ceramics terms extracted from an electronic corpus. Efficiency
and usefulness in linguistic mediation), funded by the Generalitat Valenciana (project code: GV05/260).
2
“ONTODIC. Metodología y tecnologías para la elaboración de diccionarios onomasiológicos basados en
ontologías. Recursos terminológicos para la e-traducción” (ONTODIC. Methods and technology for the
elaboration of ontology-based onomasiological dictionaries. Terminological resources for e-translation),
funded by the Spanish Ministry of Science and Technology (project code: TSI2006-01911).
TIA’09
2 The conceptual nature of activities
In this section we analyse the nature of the conceptual class activity as described
by Sager & Kageura (1994), whose model was used in the elaboration of the
catalogue of conceptual relationships developed and described in detail in Maroto
(2007).
Then we consider the characteristics of the so-called “dynamic concepts” defined
by Pilke (2001) and taken up again for the analysis of conceptual relations in
processes by Nuopponen (2007).
The concepts linked through a relationship are assigned to a conceptual class. In
their conceptual model, Sager and Kageura (1994) identify four types of concepts:
entities, activities, properties and relationships. In the following paragraphs we
present Sager and Kageura’s definition of each type of concept, supplying examples
from our empirical analysis of the field of industrial ceramics:
• a) Entity: A type of concept obtained by the abstraction of elements
from experience and reflection whose existence is considered to be independent in space
and time. Entities can be defined separately and are necessary to identify and classify the
units of experience and knowledge. All ceramic products (wall tiles, floor tiles), raw
materials (clay, stoneware), machinery and their components (oil hydraulic press, single-
deck roller kiln), as well as the places where processes occur (dryer) and where products
are used (walls, floors) would be examples in the field of industrial ceramics.
• b) Activity: A type of concept obtained by the abstraction of processes,
operations or events performed by or with entities. Their structure is more complex than
that of entities because they can only be carried out with the direct participation of the
latter. Some of the activities identified in the field of ceramics are manufacturing
processes (dust pressing, firing) and tile-laying processes (tiling, thin-set tiling).
• c) Property: A type of concept derived from the analysis of the
components and characteristics of entities, activities and relationships. Properties are
always considered to be associated to other concepts on a first level of abstraction, and
they are only constituted as independent concepts on a second level. They allow for the
identification of the differences between entities and activities, while they also reflect
their features and characteristics. Some examples in our thematic area would be all the
characteristics of ceramic products (frost resistance, porosity, colour).
• d) Relationship: A type of concept obtained from the abstraction of
physical and temporal relationships or other types of ontological relationships among
objects, and from the logical relationships among entities, relationships and activities.
Relationships are the type of concepts that identify the links that exist or have been
established between two or more entities, activities or properties, or any combination of
the three. Some examples of relationship concepts identified in the field of ceramics are
phase (indicates a sequential relationship) and composition (indicates a meronymic
relationship).
According to this classification of conceptual classes, industrial processes such as
firing belong to the conceptual class activity. This conceptual class is complex, as it
Conceptual relations established by processes
necessarily involves the direct participation of entities. That is why in ontological
engineering they are sometimes described as secondary concepts.
Nuopponen (2007: 201) describes a typical process as “a whole consisting of a
series of actions or operations, and having a start and a finish”. Processes consist of a
designed set of operations which are carried out in a particular order, using particular
tools and which produce some change in the properties of entities (or even new
entities) as an outcome. Processes are therefore “dynamic concepts” in the sense
proposed by Pilke (2001: 239), who defines them as concepts that can be realized
either as an action or an event belonging to a certain specialist field.
Based upon Pilke’s work, Nuopponen (2007: 210) applies her typology of
relations (Nuopponen, 2005) to the description of the Japanese tea ceremony, and
identifies five broad types of relations (with their subtypes) which are particularly
relevant when describing activities, actions or processes, which are grouped as
follows:
concept relations of contiguity, such as locative or temporal relations
activity relations, such as agent, object or teleological relations
origination relations, such as ingredient and resultative relations
instrumental relations, such as tool relations
transmission relations, such as source and target relations or sequential
transmission relations.
Most of these had already been described in her exhaustive classification of
conceptual relations (Nuopponen, 2005), and answer to the following basic questions:
Who? (agent)
What? (patient)
With what? (instrument)
How? (method)
Why? (cause)
Where? (place)
From where? (place of origin)
To where? (destination)
Through what? (intermediary route)
In our analysis we will concentrate on what we have called argumental and
circumstantial relationships (Maroto & Alcina, 2009: 246-247), whereas temporal
and place relations have not been considered at this stage of the research. Therefore,
we will try to answer the following questions: who?, what?, with what?, why? and
how?, adapting them to our own catalogue of relationships.
3 Conceptual relationships considered in the analysis
TIA’09
The catalogue of conceptual relations used in this study was developed as part of
our previous research (Maroto, 2007 and Maroto & Alcina, 2009). An exhaustive
description of the catalogue exceeds the scope of this paper, but we will nevertheless
devote this section to summarize what we understand by conceptual relation and the
main groups that have been established, with special emphasis on argumental and
circumstantial relations, in which processes are always involved, and therefore are
likely to appear when we try to represent knowledge about processes such as
“cocción” (firing).
We consider relationships as semantic links between two or more specialized
concepts. This definition, put forward by Otman (1996), is expressed by means of the
notation a R b, where a and b are concepts linked by the relationship R.
In the notation a R b, concepts a and b linked by R belong to one of the conceptual
classes described above (entities, activities or properties), while R has certain
properties (transitivity, symmetry, cardinality and the existence of an inverse relation)
that have consequences in the formal representation of each relationship.
Each relationship has been designated in such a way that the name shows the role
played by the concepts linked, that is, which element accounts for concept a (usually
called domain of the relationship) in the notation, and which element is represented
by concept b (usually called range of the relationship). This naming convention –
already used by Sager (1990) to name complex relationships– reveals clearly the
nature and directionality of the relationship, while making it possible to identify the
conceptual classes to which the domain and range of a relationship belong, and
allows to test whether it is suitable for the description of the relations established by
processes (activities).
Five groups of conceptual relationships were established in the catalogue: logical
relationships, meronymic relationships, sequential relationships, argumental and
circumstantial relationships, and other relationships3.
In this article we focus on argumental and circumstantial relationships. The reason
why we have chosen this group of relationships is that one of the elements involved is
always a process, and therefore belongs to the conceptual class activity. The
association of argumental and circumstantial relationships is based on a proposal by
Dancette and L’Homme (2004), which coincides with the relationships that Sager
(1990) calls complex relationships. They can be defined as the paradigmatic
conceptual relationships that are established between predicates and their arguments
(argumental relationships), as well as those which indicate the circumstances in
which a predicate occurs (circumstantial relationships). These relations give answer
to some of the above-mentioned questions suggested by Nuopponen (2007).
The following argumental and circumstantial relationships are included in our
catalogue: process–agent, process–product, process–patient, process–instrument,
process–state, process–method, cause–effect and object–use. We will now define
briefly each relationship, explicitly stating the conceptual classes that they link and
the question that they answer.
3
For a more comprehensive description of each relationship, see Maroto (2007) and Maroto and Alcina
(2009).
Conceptual relations established by processes
Process–agent: Relationship established between a process and the entity or
entities that carry out the process. These entities can be both animate and
inanimate. The conceptual classes involved can be activities (processes) and
entities (agent). This relationship answers the question Who carries out the
process?
Process–product: Relationship established between a process and the final
product resulting from the process. The conceptual classes involved can be
activities (processes) and entities (products). This relationship answers the
question What entity or entities result from the process?
Process–patient: Relationship established between a process and the entity on
which the process is carried out. The conceptual classes involved are activities
(processes) and entities (patient). This relationship answers the question On what
entity is the process carried out?
Process–instrument: Relationship established between a process and the
instrument used to carry out the process. The conceptual classes involved can be
activities (processes) and entities (instruments). This relationship answers the
question With what entity is the process carried out?
Process–state: Relationship established between a process and the final state of
the patient of the process. The conceptual classes involved can be activities
(processes) and properties (state). This relationship answers the question What is
the final outcome of the process? or even How does the process change the prior
state of affairs?
Process–method: Relationship established between a process and the method
used to carry it out. The conceptual classes involved are always activities. This
relationship answers the question How is the process carried out?
Cause–effect: Relationship established between a cause and the effect it
produces. The cause and effect can be entities, activities or properties. This
relationship answers the question Why is the process carried out? or What is the
intention or purpose of the process?
Object–use: Relationship established between an object and the use it is meant
for. This relationship can be established between entities (object) and properties
or activities (use). This relationship answers the question What is the object
(final outcome of the process) used for?
As can be seen, in every relationship at least one of the concepts involved (either
the domain or the range, or both) belongs to the conceptual class activity.
4 Empirical analysis: the case of “cocción”
Firing (“cocción” in Spanish) is the crucial stage in the production of ceramic
products. It consists of a set of operations –carried out in different kilns that cause
TIA’09
certain physical and chemical reactions that confer the final product special
properties.
We have chosen this concept as a starting point for the analysis of the relationships
established by processes because it is central in the ceramic industry.
Both a manual analysis of an introductory text about firing and the analysis of
concordances of the term “cocción” in a corpus of specialized texts were combined in
order to extract the concepts related to “cocción” by an argumental or circumstantial
relation. We have undertaken these two analyses because, although a representative
corpus yields a larger set of examples, the study of an introductory text allows the
researcher to reach a better understanding of the whole process of firing, which will
later enable to establish relationships in the corpus.
The introductory text chosen for the manual analysis is a chapter devoted to firing
in a 2005 handbook about the technology of ceramic materials written originally in
Spanish by an expert in the field (Morales Güeto, 2005).
For the analysis of concordances of the term “cocción” we have explored the
TXTCERAM corpus, which is a monolingual specialized corpus in Spanish about
ceramic industry made up by 34 specialized texts which contains 2,444,791 words.
The term “cocción” appears 4,224 times in the corpus. We have looked at the first
300 concordances in order to get a representative idea of the concepts related to
“cocción”.
Concepts related to “cocción” by argumental and circumstantial relationships
have been identified in context, deciding in each case the type of argumental or
circumstantial relationship that links each concept to “cocción”. Table 1 contains
examples of each relationship extracted from the corpus. The first column shows the
different argumental and circumstantial relationships, whereas the second shows the
two concepts linked by the relationship (one of them being always “cocción” or a
closely related concept).
Argumental and circumstantial Concepts related (domain-range)
relationship
Process-agent Cocción-hornero (firing-kiln operator)
Cocción-horno (firing-kiln)
Process-patient Cocción-pasta cerámica (firing-ceramic
paste)
Cocción blanca-arcilla china (white
firing-Chinese-clay)
Process-product Primera cocción-bizcocho (first firing-
biscuit)
Tercera cocción-decoración (third
firing-decoration)
Process-state Cocción-estado vítreo (firing-vitreous
state)
Cause-effect Cocción-contracción (firing-
contraction)
Cocción-sinteración (firing-sintering)
Conceptual relations established by processes
Process-instrument Cocción-horno (firing-kiln)
Medición de la temperatura de cocción-
pirómetro (measure of firing
temperatura-pyrometer)
Process-method Cocción- método del microscopio
electrónico (firing-electronic
microscope method)
Cocción-gresificación (firing-
vitrification)
Object-use --
Table 1. Examples from the empirical analysis.
We have found examples of all but one of the relationships presented in the
previous section. The reason why we have not found examples for the object-use
relationship is that it refers to the link between a final product and the application it is
meant for, and we have been looking at the process of making these final products.
We believe that if we examined other activities such as the tile-lying processes, this
relationship is likely to become more relevant and yield more examples.
The assignment of a particular argumental or circumstantial relationship has not
always been free of difficulties. Some problematic issues we have encountered in the
analysis are the following:
Sometimes it is hard to differentiate the process-instrument and the process-
agent relationships. Although it is generally admitted that the agent must be
animate, in the case of the ceramic industry, most processes are carried out
without any human or other animate intervention. We can distinguish between
processes in which an animate agent operates an instrument (for example, if
someone uses a hammer in order to carry out an action, the hammer would be an
instrument and the person who operates it would be the agent), whereas the
process of firing a ceramic piece could be said to be carried out entirely by the
kiln, that could be considered the agent of the whole process, as there is hardly
human intervention. There seems to be a difference depending on the degree of
automatization of the process. This difficulty could be overcome by establishing
a clear criterion that distinguishes between those processes in which there is
human intervention, where the animate entity (normally a human being) would
be the agent and the inanimate entity used would be considered an instrument,
and those where the inanimate entity can be considered the agent of the whole
process, because it plays a quasi-active role in the process.
A second difficulty arises from the fact that the term “cocción” is polysemous.
The analysis shows that this term is used to refer both to the whole firing cycle,
and to the particular step of the cycle in which the final properties of the
finished product are achieved through physical and chemical reactions. This
difficulty could be overcome by considering both concepts separately and
establishing two separate sets of related terms depending on the meaning.
TIA’09
A third problem arises from the fact that our analysis is based only on
contextual information, which is not always enough in order to clarify the nature
of the relationship. For example, in view of the contexts in the corpus, one may
think that the relationship between “horno” (kiln) and “cocción” (firing) could
be either the following three:
- Process-place: the kiln is the physical space where the firing occurs.
Ex.: “horno” (kiln) is defined as “Una estructura en la que un material o producto se
cuece, calcina o se somete de alguna manera a temperaturas elevadas”. (A structure in
which a material or product is fired, calcined or is subject somehow to high
temperatures).
- Process-instrument: the firing is carried out using a kiln. Ex.: “A pesar
de la ingeniería que se le haya aplicado, el horno continúa siendo un instrumento
fascinante y temperamental.” (No matter how much engineering has been applied, the
kiln keeps being a fascinating and temperamental instrument).
- Process-agent: As we mentioned above, the process of firing a ceramic
piece could be said to be carried out entirely by the kiln, that could be considered the
agent of the whole process, in which there is hardly human intervention.
This kind of ambiguity could be resolved with the help of an expert in the field
who could actually decide which relationship is more suitable from an industrial point
of view, or even, it may be convenient to establish more than one relationship
between the same two concepts depending on the stage of the process or other
factors.
• Sometimes it was also difficult to distinguish between object-use and
the process-instrument relationships. What we consider an instrument is any artefact that
helps in carrying out a process, whereas the kind of object involved in the object-use
relationship is considered to be a finished product and not a means of carrying out the
action.
• Finally, we have become aware of the relevance and complexity of the
cause-effect relationship in any process. In analysing causes and effects we have noticed
that, as Nuopponen (2005) had already pointed out, there is a difference between
consequence causal relations and causal coordination of concepts. That is, we can
distinguish between single causes and effects (for example humidity causes corrosion)
and chains of causes and effects, where the effect of the first cause may be the cause of
another effect (for example humidity causes corrosion and corrosion causes the
appearance of wholes on the surface of a ceramic product). Another possibility to
approach causal relationships could be considering what Barrière (2002: 98) calls
“semantic relation refinement”, which allows to subdivide causal relationships depending
on the granularity of the analysis and therefore to establish a hierarchy of causal
relationships.
5 Implementation of the concept “cocción” in the ontology editor
Protégé
Conceptual relations established by processes
Once the concepts related to “cocción” through argumental and circumstantial
relationships have been identified, this concept and the related concepts have been
introduced in the conceptual database Ontoceram, created using the ontology editor
Protégé-frames (Stanford Medical Informatics, 2009).
In this database, concepts are represented through the Class component of
Protégé. The concept “cocción” is inserted in the concept hierarchy as a subordinate
of PROCESO (“process”), which in turn depends from the top concept ACTIVIDAD
(“activity”). The hierarchical structure enables to reflect the relationship between
generic and specific concepts. For example, “monococción” (single firing) is a
subtype of firing and therefore is situated depending from “cocción” in the hierarchy
of concepts.
The concepts related to “cocción” are also represented through the Class
component of Protégé. The link between “cocción” and its related concepts is
represented through the Slot component, which allows for the assignment of
attributes, thus making the relationships explicit in Ontoceram.
Terms are represented through the Instance component of the ontology editor,
thus making it possible to retrieve the information through queries (Maroto & Alcina,
2009: 255-256).
In the screenshot shown in Fig. 1 it can be seen how the concept Cocción and its
related concepts appear in Ontoceram. This concept is subordinated to PROCESO
(process) and has been assigned several relationship slots that link “cocción” to other
concepts also included in the database. For example, “cocción” has been related to
“contracción” (contraction) and “sinterazión” (sintering) through the slot “causa-
efecto” (cause-effect).
Fig. 1 – Implementation of the concept “cocción” (firing) in Ontoceram.
TIA’09
6 Conclusion
The analysis of the argumental and circumstantial relations established by the
process of firing in the ceramic industry presented in the contribution has allowed us
to test the validity of the catalogue of conceptual relationships established in our
previous research for the representation of knowledge about activities.
Although the relationships initially proposed seem to be useful for this task, there
are several aspects that need further refinement.
First, we need to better define the nature of the concepts linked by each
relationship in order to avoid ambiguity. Experts in the field could help in order to
determine the type of relationship that better suits the requirements of the specialized
field when the analysis of specialized texts suggests that more than one relation is
possible. We should also establish clear criteria in order to define the scope of each
argumental and circumstantial relationship.
Second, the cause-effect relationship needs to be further refined in order to better
represent the possibility of cause-effect chains.
Finally, the analysis of argumental and circumstantial relationships needs to be
complemented with the study of other relevant relationships, such as generic-specific
relationships (in order to reflect different kinds of firing processes) and meronymic
relationships (in particular the stage-process relationship). Other relevant
relationships that need further analysis are sequential relationships (both spatial and
temporal) in order to fully represent processes.
References
ALCINA, A. (2009). Metodología y tecnologías para la elaboración de diccionarios
terminológicos onomasiológicos. In ALCINA, A., E. VALERO and E. RAMBLA Eds.
Terminología y sociedad del conocimiento. Berna: Peter Lang, p. 33-58.
BARRIÈRE, C. (2002). Investigating the Causal Relation in Informative Texts. Terminology,
Vol.7, nº2, p. 135-154.
DANCETTE, J. & M.-C. L’HOMME. (2004). Building Specialized Dictionaries Using Lexical
Functions. Linguistica Antverpiensia, NS, nº 3, p. 113-131.
MAROTO, N. (2007). Las relaciones conceptuales en la terminología de los productos
cerámicos y su formalización mediante un editor de ontologías. PhD Thesis. Available at:
http://www.tesisenred.net/TDX-0306109-103431.
MAROTO, N. & ALCINA, A. (2009). Formal Description of Conceptual Relationships with a
View to Implementing Them in the Ontology Editor Protégé. Terminology, Vol.15, nº 2, p.
240-265.
NUOPPONEN, A. (2005). "Concept Relations: An Update of a Concept Relation Classification".
In: MADSEN, B. N. & H. ERDMAN Ed. Copenhagen 7th International Conference on
Terminology and Knowledge Engineering, TKE 2005. Copenhagen: Association for
Terminology and Knowledge Transfer, p. 127-38.
NUOPPONEN, A. (2007). Terminological Modelling of Processes. An Experiment. In ANTIA, B.
E. Ed. Indeterminacy in Terminology and LSP. Studies in honour of Heribert Picht.
Amsterdam/Philadelphia: John Benjamins, p. 199-216.
OTMAN, G. (1996). Les représentations sémantiques en terminologie. Paris: Masson.
Conceptual relations established by processes
PILKE, N. (2001). Field-Specific Features of Dynamic Concepts: What, When and Why? In F.
MAYER Ed. Language for Special Purposes: Perspectives for the New Millennium. Tübingen:
Gunter Narr, Vol. 1, p. 239-246.
SAGER, J.C. (1990). A Practical Course in Terminology Processing. Amsterdam/Philadelphia:
John Benjamins.
SAGER, J.C. & KAGEURA, K. (1994). Concept Classes and Conceptual Structures: Their Role
and Necessity in Terminology. Actes de Langue Française et de Linguistique (ALFA):
Terminology and Special Linguistics, Vol. 7/8, p. 191-216.
STANFORD MEDICAL INFORMATICS. (2009). The Protégé Ontology Editor and Knowledge
Acquisition. Available at: http://protege.stanford.edu/.