Domain Ontology to Support Open Data Analytics for Aquaculture
Pedro Oliveira, Ruben Costa, José Lima, João Sarraipa and Ricardo Jardim-Gonçalves
CTS, UNINOVA, Dep.º de Eng.ª Electrotécnica, Faculdade de Ciências e Tecnologia, FCT,
Universidade Nova de Lisboa, 2829-516 Caparica, Portugal
Keywords: Ontology Engineering, Knowledge Representation, Multilingual, Education and Training.
Abstract: The Aquaculture industry, which comprises mainly of SME companies, represents a significant source of
protein for people. From an IT perspective, aquaculture is characterized high volumes of heterogeneous data,
and also lack of interoperability intra and inter-organisations. Each organization uses different data represen-
tations, using its native languages and legacy classification systems to manage and organize information. The
lack of semantic interoperability that exists can be minimized, if innovative semantic techniques for repre-
senting, indexing and searching sources of non-structured information are applied. The work presented here,
describes the achievements under AQUASMART EU project, which aims to accelerate innovation in Eu-
rope’s aquaculture through technology transfer for the deployment of an open data solution through multilin-
gual data collection and analytics solutions and services, turning the large volumes of heterogeneous aqua-
culture data that is distributed across the value chain, into an open cloud of semantically interoperable data
assets and knowledge. Results achieved so far do not address the final conclusions of the project but form the
basis for the formalization of the AQUASMART semantic referential.
1 INTRODUCTION
The aquaculture industry, which comprises mainly of
SME’s companies, represents a significant source of
protein for people. Globally, nearly half the fish con-
sumed by humans is produced by fish farms. Aqua-
culture is now fully comparable to capture fisheries
when measured by volume of output on global scale.
The contribution from aquaculture to the world total
fish production of capture and aquaculture in 2012
reached 42.2 percent, up from 25.7 percent in 2000
(FAO Fisheries and Aquaculture Department, 2014).
Global production is forecasted to increase from 45
million tons in 2014 to 85 million by 2030, making
the aquaculture industry the fastest growing animal
food producing sector in the world. The European
Union needs an innovative aquaculture industry to
meet rising seafood demand and to enhance its com-
mercial stocks.
According to the Food and Agriculture Organiza-
tion of the United Nations (FAO), the volume and
value data in global aquaculture production are pri-
marily official statistics obtained directly from the na-
tions and mainly described in local language. Also,
there are available other relevant sources of data, like
academic reviews, consultant reports and other spe-
cialist literature.
The main problems of the Aquaculture sector are
related with the lack of global knowledge access, and
the inefficient data exchanges and data reuse between
aquaculture companies and its related stakeholders.
This is primarily due to incompatibility problems
among the several information representation struc-
tures used by the different software applications along
supply chains and business networks (Ray and Jones,
2006). Aquaculture companies have limited capabili-
ties to hire specialized technical resources out of their
core business. The main issue fish farmers’ face is
data understanding and identifying correlations be-
tween parameters that affect production, the lack of
skilled professionals and the right IT tools, prevents
fish farmers to get better insights of their own data
and also prevents them to share best practices with
other aquaculture stakeholders. For example, if one
could reach other growers data (e.g., growth rates,
FCR (Feed Conversion Ratio) – related to environ-
mental conditions of cause), than it would be able to
have a better and closer prediction plus better and
closer feeding according to the current biomass. So,
an important step is to be able to get actionable in-
sights in the data resulting in smarter decisions and
344
Oliveira, P., Costa, R., Lima, J., Sarraipa, J. and Jardim-Gonçalves, R..
Domain Ontology to Support Open Data Analytics for Aquaculture.
In Proceedings of the 7th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2015) - Volume 2: KEOD, pages 344-351
ISBN: 978-989-758-158-8
Copyright
c
2015 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
better business outcomes, being able to look at past
performance and understand that performance by
mining the related data (production, environment,
etc.) to look for the reasons behind past success or
failure and take better decisions for the future.
Knowledge transfer goals are to take the state of
the art in multilingual data collection tools, analytics
solutions and services, semantic interoperability
methods and data mining procedures to implement a
global open aquaculture data information system,
able to respond to specific user needs and profiles.
Users in the Aquaculture industry can innovate
taking the novel capabilities for seamless and holistic
access of multilingual data products and services in
the Aquaculture value chain, bridging across borders,
languages, industries and sectors, removing barriers
both technical and organizational.
The access of this global knowledge creates sig-
nificant new and further commercial opportunities
and positive economic prospects.
In India several ICT initiatives are being tried for
disseminating aquaculture information to fish farm-
ers, shortening the digital gap and helping the farmer
in reaping a good harvest.
There is a demand for intelligent world-class so-
lutions capable of reinforcing partnerships and col-
laborations with an improved cross-cultural under-
standing. However due to the proliferation of termi-
nology, organizations from similar business environ-
ments have trouble cooperating, and are experiencing
difficulties exchanging electronically vital infor-
mation. To address these issues it is essential to de-
velop semantic tools that easy these integration pro-
cesses.
Some examples of closely related research
streams in recent years are: the extensive work on
knowledge models and knowledge management
tools, the rise of so-called knowledge engineering, the
myriad of projects around ‘controlled vocabularies’
(such as ontologies, taxonomies, dictionaries, and
thesauri), and the academic knowledge-centred
courses (graduation, master, and doctoral).
A controlled vocabulary is a list of terms that have
been enumerated explicitly. This list is controlled by
and is available from a controlled vocabulary regis-
tration authority. All terms in a controlled vocabulary
should have an unambiguous, non-redundant defini-
tion. This is a design goal that may not be true in prac-
tice. It depends on how strict the controlled vocabu-
lary registration authority is regarding registration of
terms into a controlled vocabulary. At a minimum,
the following two rules should be enforced (Lima et
al., 2007):
If the same term is commonly used to mean dif-
ferent concepts in different contexts, then its name
is explicitly qualified to resolve this ambiguity.
If multiple terms are used to mean the same thing,
one of the terms is identified as the preferred term
in the controlled vocabulary and the other terms
are listed as synonyms or aliases.
In line in the notion of controlled vocabularies, an on-
tology for the Aquaculture domain is being devel-
oped. The approach presented here will be beneficial
for aquaculture companies in several ways: (i) by en-
abling to benchmark similar companies with relation
to their production performance indicators; (ii)
providing access to consolidated information on best
practices and success stories; (iii) to be able to inter-
pret their production data using data mining tech-
niques and share that data through the
AQUASMART open cloud.
This paper is structured as follows: Section 2 pre-
sents the related semantic approaches. The overview
and architecture of the project that serves as valida-
tion for the ontology are explained in section 3. Sec-
tion 4 presents the AQUASMART semantic referen-
tial. Section 5 presents the IT semantic services de-
veloped, the tools used and the interdependencies be-
tween components that implement the
AQUASMART functionalities. Finally, section 6
presents our main conclusions and additional consid-
eration on future work.
2 RELATED WORK
Big Data analytics and knowledge extraction from the
aquaculture domain, is something relatively new.
This chapter address some relevant works that have
been done relatively to the technology and the do-
main.
2.1 ICT in Aquaculture
Recent ICT projects on the aquaculture domain, aim
to ease the access to information about scientific
farming practices and market prices through web por-
tals (De et al., 2008), while others propose innovative
and unique models of information exchange on farm
history, crop details, soil details, weather data, farmer
details, case sheets, photo bank and a library (Vimala,
2009). The multilingual issue is addressed by (Mon-
dal et al., 2011), where an online tool provides an-
swers to questions asked by farmers and agro-profes-
sionals over the Internet.
Also, a very important step to adopt ICTs in Aq-
uaculture environments is to reach the end-users in an
Domain Ontology to Support Open Data Analytics for Aquaculture
345
early-stage, even during the learning process. In
(Seixas et al., 2015) the authors review educational
means used in teaching and learning in the area of aq-
uaculture, fisheries and aquatic resources manage-
ment at European level, with specific consideration
on the use of ICT and e-learning tools. It concludes
that there is a real and urgent need to "train the train-
ers" to use ICT in their teaching environments. Addi-
tionally, from the students' end, there is a strong de-
sire to learn more about the application of e-learning
tools and to use them in their learning process.
2.2 Ontology Development
The method proposed in our work, adopts an associ-
ation rules learning technique in order to discover rel-
evant relations among key terms in a document cor-
pus, and additional human input to perform the map-
pings between terms (frequent item sets) and ontolog-
ical concepts and the establishment of the final scores
on each relation. In simple words, frequent item sets
are groups of items that often appear together in the
data.
This will reuse the findings proposed by (Paiva et
al., 2013) and extend it to the aquaculture domain.
Figure 1 depicts the methodology to be addressed by
this work regarding thesaurus building.
Figure 1: Thesaurus building methodology (Sarraipa et al.,
2010).
One of the focus of the work addressed where, is
the development of a semantic referential for the aq-
uaculture domain. Regarding ontology development,
the scientific literature is vast and addresses several
approaches and tools for dealing with its specification
and management. Several efforts have focused on ex-
tracting structured ontologies from unstructured text.
2.3 Multilingual Data
The introduction of an innovative multilingual
knowledge base capacity suitable for the Aquaculture
sector, which would enable large volumes of data to
be accessible as semantic interoperable data and
knowledge will improve significantly the sector and
ultimately the EU’s competitiveness.
Three tools to support the multi-linguist require-
ments: 1) a tool for language independent document
representations (Canonical Correlation Analysis
(CCA) mappings) (Kuss and Graepel, 2003); 2) a tool
for multilingual semantic annotation (Enrycher)
(Jožef Stefan Institute, 2012); and 3) a statistical ma-
chine translation tool (MOSES) (Koehn et al., 2007).
They will be used to support the ontology mapping
processes which involve ontology concepts and anno-
tated documents in several languages.
Enrycher is a service-oriented system providing
shallow as well as deep multilingual text processing
functionality at the document level. It can be used to
perform multilingual semantic annotation on docu-
ments in common open knowledge bases such as
DBpedia (Lehmann et al., 2015), and can support the
local to reference ontology mapping processes.
The MOSES is an open source statistical machine
translation system to translate documents stored in lo-
cal databases across several language pairs in order to
share the knowledge between Aquaculture compa-
nies. The system will also be used to support the mul-
tilingual training and learning platform by translating
training materials such as best practices documents
and video lectures. Linguistic resources such as the
Fisheries Glossary can be incorporated in the process
(Countryside Council for Wales, 2001) by using it as
a parallel corpus. Having large collections of short
documents can be used to adapt the language model
for the target language.
3 AQUASMART CONCEPT
The data collected in the AQUASMART open data
cloud is suitable to be reused in other industrial do-
mains if needed, (e.g., environmental or transporta-
tion data), providing a cross-sectorial setting to the
provided solution. The cloud is enriched with a layer
of multilingual information (multilingual mappings
and multilingual linguistic information) and with a set
of services for creating, representing and accessing
that multilingual information.
The prime goal is to accelerate innovation in Eu-
rope’s Aquaculture through technology transfer for
KEOD 2015 - 7th International Conference on Knowledge Engineering and Ontology Development
346
the deployment of an open data solution through mul-
tilingual data collection and analytics solutions and
services, turning the large volumes of heterogeneous
aquaculture data that is distributed across the value
chain, into an open cloud of semantically interopera-
ble data assets and knowledge. Each such systems
usually uses different data representations, using its
native languages and knowledge organization tools
such as vocabularies and classification systems to
manage and organize information. Although the prac-
tice of using knowledge organization tools to support
document tagging (e.g. thesaurus-based indexing)
and information retrieval (e.g. thesaurus-based
search) improves the functions of a particular infor-
mation system, it is leading to the problem of integrat-
ing information from different sources due to lack of
semantic interoperability that exists among
knowledge organization tools used in different infor-
mation systems. The project technology transfer goals
are to take the state of the art in multilingual data col-
lection tools, analytics solutions and services, seman-
tic interoperability methods and data mining proce-
dures to implement a global open aquaculture data in-
formation system, able to respond to specific user
needs and profiles Figure 2.
Figure 2: AQUASMART Concept.
Actually, aquaculture companies have never tried
to transform data that are captured into knowledge,
and share this knowledge to improve efficiency, in-
crease profitability and do business in a sustainable,
environmentally friendly way. In other words, data
are just captured but not exploited. AQUASMART
represents a big innovation in this direction, it adds
the dimension of global open data. The improvement
and the innovation become even bigger because the
quality of the analysis that will be performed by the
companies will be dramatically improved. Hence,
even the exchange of information itself is a huge busi-
ness innovation. It is the first time that companies will
be able compare their results to the ones of other com-
panies and benchmark their performance. This is go-
ing to create a multiplier effect and boost competition
for improvement in the sector. This will be further fa-
cilitated and enhanced by the integration with the Eu-
ropean Union Open Data Portal and the exchange of
information in the social media.
AQUASMART vision relates to implementing a
state of art multilingual open data framework that
companies can use to seamlessly access global data
and take more knowledgeable decisions using multi-
lingual information. H2020’s vision suggests that en-
terprises must move away from silo solutions, used
behind the closed doors of company, to a more open
data technological solutions built for the industrial
sector to enhance their operations. However, the ac-
tual state of practice is that knowledge is transferrable
and sharable but with significant barriers for semantic
compatibility.
The main mission is driven by the business need
of the European aquaculture companies, when com-
panies have business objectives that they cannot
achieve due to lack of instruments that would enable
them to manage and access to global knowledge and
big data, in a multi-lingual, multi-sector and cross-
border setting.
4 SEMANTIC REFERNCIAL
To improve the ability of aquaculture companies to
innovate across their value chain, there is a need to
provide multi-lingual data, which must be interoper-
able through various products or services. To achieve
this objective, the proposed approach integrates a ref-
erence ontology that supports the integration of het-
erogeneous data (including multilingual) that con-
cerns the aquaculture environment.
The proposed semantic referential provides the
distributed composition, formality, richness and qual-
ity of information required among the aquaculture
sector to ensure that all the actors within the produc-
tion process “speak the same language”.
The creation of the semantic referential followed
a method for designing and developing a domain on-
tology with inputs from knowledge experts, provid-
ing the necessary insights towards the improvement
of the efficiency of the aquaculture production pro-
cesses. Such experts, contributed with their
knowledge about the aquaculture production, the ac-
tors involved, and the data generated during the pro-
duction process (Oliveira et al., 2015).
Domain Ontology to Support Open Data Analytics for Aquaculture
347
Figure 3: AQUASMART Semantic Referential.
This semantic referential, as seen in Figure 3, is
composed by four main areas, (i) the AQUASMART
production domain; (ii) FAO aquaculture glossary;
(iii) AQUASMART training modules and; (iv) the
AQUASMART IT framework. In the following sec-
tions, the authors address each of these modules,
which comprises the overall AQUASMART seman-
tic referential. Each of the modules are being inte-
grated under the same umbrella (AQUASMART se-
mantic referential) and formalized in OWL language.
4.1 AQUASMART Production Domain
In the AQUASMART context, knowledge experts are
the end users (mainly fish farmers). The purpose of
involving such experts in the process, is not only to
provide input to the semantic referential, but also to
perform a quality review of the AQUASMART train-
ing courses. With the help of these experts, the main
structure of the AQUASMART ontology was devel-
oped to accommodate all the important and necessary
information that will support all the project services
and functionalities.
The ontology is mainly separated in two concepts,
the “Aquaculture Production Entities” and the “Grow
Out Data Analysis”. The first contains the all the aq-
uaculture related entities, while the second one con-
tains the key performance, the process and production
related data.
The “Aquaculture Production Entities”, are the
main components of the production operation, includ-
ing actors involved in the process, species being pro-
duced (e.g. seabream and seabass), location of the fish
farm, and the type of cages used to store the fish. The
“Grow Out Data Analysis”, is focusing on process
steps, production data and indictors. A sample of rel-
evant production parameters is shown on Table 1.
Table 1: Production Parameters.
Geographical Region Species
Hatchery Broodstock origin
Hatchery Quality (text) Stocking Month
Average Weight Mortality
Avg. Temperature Avg. SFR
Avg. Fish Density Oxygen
This ontology will be integrated with an aquacul-
ture glossary that contains the aquaculture related
terms and their definitions. To make available fully
interoperable multi-lingual data products and services
in the Aquaculture, AQUASMART makes use of the
FAO glossary for aquaculture (FAO, 2015).
4.2 FAO Aquaculture Glossary
From the AQUASMART perspective, the multilin-
gual data generated within the aquaculture domain,
can be exploited as a layer of services and resources
by seamlessly adding (i) linguistic information for
data and vocabularies in different languages, (ii) map-
pings between data with labels in different languages,
and (iii) services to dynamically access and traverse
linked data across different languages.
We envisage a multilingual aquaculture where an
end-user would query the “Aquaculture Open Data
Cloud” in his/her own language, and would get the
relevant data in that language.
The primary objectives of the AQUASMART
glossary is: (i) to serve as a reference to fish farmers,
consultants, administrators, policy makers, develop-
ers, engineers, agriculturists, economists, environ-
mentalists and any other actor interested in aquacul-
ture; (ii) and to facilitate communication among ex-
perts and scientists involved in aquaculture research
and development.
Figure 4: Glossary Term Example.
The glossary supports a multi-lingual approach
that includes, in an initial phase, English, French,
KEOD 2015 - 7th International Conference on Knowledge Engineering and Ontology Development
348
Greek, Hebraic, Spanish and Portuguese terms. Each
term has properties which define it (Name, definition,
related term, synonyms, subject area, translation, and
image). This kind of information is what defines and
gives semantic meaning to terms. Figure 4 shows an
example of a glossary term.
4.3 Ontology Training
The training development methodology translates de-
sign specifications into training materials. The meth-
odology presented by Sarraipa et al., (2013) starts by
identifying the training objectives and the target au-
dience including desired roles & competences. Then
it uses an appropriate instructional approach to per-
form the training courses’ materials development,
complemented with a set of different quality reviews.
The overall process of developing training fol-
lows a specific process, composed by three different
task tracks (training development, overall training
validation and training execution) that complement
each other. The training development track starts by
defining the course’s synopsis according to the direc-
tives obtained from a training overall objective. Thus,
it was identified the need of defining all these objec-
tives in a courses synopsis. A course synopsis is an
official description of the course as stated in the insti-
tution's catalogue of courses. It should indicate the
overall goal of the course, briefly characterize the
main topics covered, point out why the course is im-
portant to students, identifying any special instruc-
tional methods to be used, and comment on what
background students should have in order to best ap-
preciate the course content. The courses synopses
also act as guidelines to the training course’s authors.
Figure 5: Training Ontology.
The AQUASMART ontology will also be used, to
represent the training KB (Knowledge Base), facili-
tating the categorization of its elements and subse-
quently reasoning over it. It should contribute to the
skills and competencies development of the trainees
as required for specific understanding and exploita-
tion. This reflects the need to develop, organise and
run courses, for example, to train “future users”, in
how to use a specific software or extracting relevant
patterns from aquaculture production data. Figure 5
presents the relations between training concepts built
in Protégé (Stanford Center for Biomedical Informat-
ics Research, 2012).
In this model each learning Module has several
concepts associated, the Sources concept contains in-
formation about the sources referred in the Module,
Contact includes the contact information of the author
of a Module or Course and Keywords that contain a
list of all relevant keywords needed for describing the
contents of the Module. A Course, other than Con-
tacts and Modules that contain the course also in-
cludes Keywords (that include Keywords inherited
from its Modules) and belongs to a Curriculum Main
Area that is divided by Content Areas and Learning
Levels. Each Module and Course has a Target Audi-
ence Group and a Target Audience Industry, to be
recommended accordingly to the profile of the
learner. Finally, a pre-defined Programme is defined
for a specific Target Audience Industry and Target
Audience Group.
5 AQUASMART SERVICES
The AQUASMART IT framework is not yet devel-
oped at the current stage of the project. It will be in
this framework that the AQUASMART services will
persist.
One of the services is the ability to search for Aq-
uaculture partners through the ontology. The ontol-
ogy will support a search feature that allow users to
find suitable companies in Aquaculture domain to
partner. This feature allow to search companies by
different criteria like water temperature, type of fish
produced, size of production, country, this are some
examples supported.
Providing the achieved results to other users is a
service that is being worked on. The main idea of this
service is to provide the results obtained during the
project. This can be in text, formulas or other type of
format that can be understand by users external to the
project.
5.1 Knowledge Search Engine
The other service developed is the training search en-
gine to support the Aquaculture domain. Training ma-
terials are a necessary part of any program or activity
that involves knowledge acquisition and retention
(Wikihow, 2015). For this, authors found appropri-
ated to define “oLEARCH - Ontologies LEArn by
Domain Ontology to Support Open Data Analytics for Aquaculture
349
seaRCHing”, as a new concept related to ontologies
able to change/adapt their knowledge (to learn)
through their users’ patterns of searching/reasoning.
The concept was inspired from the concept LEARCH
defined by (Ratliff et al., 2009) that means “LEArn-
ing to seaRCH” and was defined to represent algo-
rithms for imitation learning in robotics with the main
purpose to search something.
oLEARCH is a training materials search engine
application available to users by Internet. This system
learns from user’s searched training materials con-
cepts improving the KB.
The oLEARCH function uses an algorithm sup-
ported in an instance-based learning approach based
on user interactions. In instance-based learning, train-
ing examples are stored verbatim, and a distance
function is used to determine which member of the
training set is closest to an unknown test instance
(Witten and Frank, 2005). In oLEARCH, such dis-
tance function is represented by the semantic dis-
tance, which is the inverse of the semantic relatedness
between the users introduced concepts and the train-
ing materials classified in the reference ontology.
Thus, oLEARCH provides to the users a set of train-
ing materials that are close to their introduced con-
cepts in terms of semantic relatedness. Then, users are
able to select the most appropriated training materi-
als from this set of possible choices. These last users’
selections are also used, as a last feedback, to increase
the semantic relatedness weight of the selected train-
ing materials associated concepts.
6 CONCLUSIONS
The work described within this paper relates to the
development of a domain ontology, able to support
and describe the analysis of aquaculture production
data, but also, the training and IT services which com-
poses the AQUASMART platform. Although final
conclusions are not yet validated, preliminary analy-
sis led us to conclude that the Aquaculture domain is
lacking for semantic approaches which enable data
understanding intra and inter-organizations. The for-
malization and validation of a common semantic ref-
erence model, which is able to drive new and dynamic
collaborations between aquaculture companies and
consequently generate new business opportunities to
them, can be seen as first step towards semantic in-
teroperability. From an application scenario perspec-
tive, the objective of the AQUASMART semantic
referential will enable the understanding of data ana-
lytics resulting from the production data. The idea is
to semantically annotate the results of the correlations
found in batches of production data, with ontology
concepts in order to give meaning to data analytics
results. Multilingual is another important feature due
to the fact that knowledge transfer is one of the main
challenges to be addressed here.
With the proposed approach presented here, there
will be opportunity for innovation in the aquaculture
industry such as transforming data into global
knowledge, and use this knowledge to improve effi-
ciency, increase profitability and do business in a sus-
tainable, environmentally friendly way; Better and
perfect view of the life to date fish behaviour and the
living inventory (biomass) that exist in a farm, based
on the analysing of all environmental and biological
data that will exist in the local system and at global
level. By knowing the global parameters that affect
the production, the companies will be able to make
accurate estimations of the growth of the fish and the
result of the production every day.
ACKNOWLEDGEMENTS
The authors acknowledge the European Commission
for its support and partial funding and the partners of
the research project: H2020-644715 AQUASMART.
REFERENCES
Countryside Council for Wales, 2001. A glossary of Marine
Nature Conservation and Fisheries. [Online]
Available at: http://jncc.defra.gov.uk/pdf/glossary.pdf
De, H. K., Saha, G. S., Srichandan, R. & Vipinkumar, V.
P., 2008. New initiatives in fisheries extension. Aqua-
culture Asia, Volume 13, pp. 16-19.
FAO Fisheries and Aquaculture Department, 2014. FAO
Global Aquaculture Production Volume and Value Sta-
tistics Database Updated to 2012, s.l.: s.n.
FAO, 2015. FAO Glossary. [Online] Available at: www.
fao.org/faoterm/collection/aquaculture/en/[Accessed
2015].
Jožef Stefan Institute, 2012. Enrycher. [Online]
Available at: http://enrycher.ijs.si/
Koehn, P. et al., 2007. Moses: Open Source Toolkit for Sta-
tistical Machine Translation. Prague, Annual Meeting
of the Association for Computational Linguistics
(ACL).
Kuss, M. & Graepel, T., 2003. The Geometry Of Kernel Ca-
nonical Correlation Analysis, s.l.: Max Planck Institute
for Biological Cybernetics.
Lehmann, J. et al., 2015. DBpedia – A Large-scale, Multi-
lingual Knowledge Base Extracted from Wikipedia. Se-
mantic Web Journal, 6(2), p. 167–195.
Lima, C., Zarli, A. & Storer, G., 2007. Controlled Vocabu-
laries in the European Construction Sector: Evolution,
KEOD 2015 - 7th International Conference on Knowledge Engineering and Ontology Development
350
Current Developments, and Future Trends. In: Complex
Systems Concurrent Engineering. s.l.:Springer London,
pp. 565-574.
Mondal, P. P., De, H. K., Saha, G. S. & Radheyshyam,
2011. Information and Communication Technology
(ICT) and Aquaculture. Aquaculture International, pp.
32-35.
Oliveira, P. et al., 2015. A Knowledge-Based Approach for
Supporting Aquaculture Data Analysis Proficiency.
Houston, Texas, USA, s.n.
Paiva, L., Costa, R., Figueiras, P. & Lima, C., 2013. Dis-
covering Semantic Relations from Unstructured Data
for Ontology Enrichment - Association rules based ap-
proach. Lisbon, IEEE.
Ratliff, N., Silver, D. & Bagnell, J. A., 2009. Learning to
search: Functional gradient techniques for imitation
learning. Autonomous Robots, 27(1), pp. 25-53.
S. R. Ray, A. T. Jones, 2006. Manufacturing interoperabil-
ity. Journal of Intelligent Manufacturing, Volume 17,
pp. 681-688.
Sarraipa, J. et al., 2013. E-TRAINING Development Ap-
proach for Enterprise Knowledge Evolution. San Di-
ego, CA, USA, s.n.
Seixas, S., Dove, C., Ueberschar, B. & Bostock, J., 2015.
Evaluation on the use of e-learning tools to support
teaching and learning in aquaculture and aquatic sci-
ences education. Aquaculture International, pp. 825-
841.
Stanford Center for Biomedical Informatics Research,
2012. Stanford's Protégé Home Page. [Online] Availa-
ble at: http://protege.stanford.edu/[Accessed 3 Septem-
ber 2012].
Vimala, D., R., T. M. P., A. K., 2009. An innovative infor-
mation and Communication technology for transfer of
technology for aquaculture. Aquaculture Asia, Volume
14, pp. 23-24.
Wikihow, 2015. http://www.wikihow.com/Develop-Train-
ing-Materials. [Online] Available at: http://www.wiki
how.com/Develop-Training-Materials[Accessed
2015].
Witten, I. H. & Frank, E., 2005. Data mining: practical ma-
chine learning tools and techniques. 2nd ed. S.l.:Mor-
gan Kaufmann Series in Data Management Systems.
Domain Ontology to Support Open Data Analytics for Aquaculture
351