=Paper=
{{Paper
|id=Vol-2228/short2
|storemode=property
|title=Building an Ontology-Based Infrastructure to Support Environmental Quality Research: First Steps
|pdfUrl=https://ceur-ws.org/Vol-2228/short2.pdf
|volume=Vol-2228
|authors=Patricia M. C. Campos,Cássio C. Reginato,João Paulo A. Almeida,Julio Cesar Nardi,Monalessa P. Barcellos,Renata S. S. Guizzardi
|dblpUrl=https://dblp.org/rec/conf/ontobras/CamposRANBG18
}}
==Building an Ontology-Based Infrastructure to Support Environmental Quality Research: First Steps==
https://ceur-ws.org/Vol-2228/short2.pdf
Building an Ontology Network to Support
Environmental Quality Research: First Steps
Patricia M. C. Campos, Cássio C. Reginato, João Paulo A. Almeida, Julio Cesar
Nardi, Monalessa P. Barcellos, Ricardo de Almeida Falbo, Renata S. S. Guizzardi
Ontology and Conceptual Modelling Research Group (NEMO) – Computer Science
Department, Federal University of Espírito Santo (UFES), Vitória, Brazil
patmarcal@gmail.com, cassio@inf.ufes.br, jpalmeida@ieee.org,
julionardi@ifes.edu.br, {monalessa, falbo, rguizzardi}@inf.ufes.br
Abstract. Environmental Quality Research (EQR) comprises many different
methods, procedures and subdomains, often requiring the integration of
heterogenous data from many sources. In this paper we show the first steps in
building an ontology network to support interoperability of EQR data. We
present a bottom-up approach that begins by analyzing available data to
capture relevant community concerns and then establish the domain coverage.
We focus on identifying and reusing existent knowledge sources to cover the
semantics of extant data.
1. Introduction
Environmental Quality Research (EQR) performs observations and measurements to
evaluate environmental conditions (Cox, 2017). This type of research usually has a
specific purpose, like analyzing water quality at a particular site or understanding how an
environmental disaster affects the quality of an ecosystem. In EQR practice, researchers
make joint use of the data they produce with existing data from many sources, such as
academic literature, environmental agencies, independent experts and consultants, etc.
The requirements of data correlation bring up serious interoperability concerns, since
each source may use different vocabularies and standards.
The FAIR Data initiative (FORCE11, 2016) presents a number of principles to
address this problem. One of them establishes that metadata must meet domain-relevant
community standards. This means that the produced data must be annotated with metadata
that, in turn, must reference domain-relevant community standards such as reference
ontologies. For several decades it has been often suggested that ontologies could be a key
instrument to address semantic interoperability challenges.
In some success cases like (Ashburner et al., 2000), ontology-based models have
become reference models that are used and reused in a large community, with beneficial
consequences on data reuse. In other cases, however, ontologies fail in establishing de
facto shareability (i.e., community acceptance) and consequently fail to support
interoperability. This may have a number of reasons, including: (i) the lack of alignment
between the various data sources and available ontologies, (ii) the coexistence of
competing and incompatible data schemas and vocabularies, (iii) the lack of consideration
of existing knowledge sources for the construction of ontologies.
Aware of such risks, we are currently investigating an ontology development
bottom-up approach that can cope with the demands of data integration by leveraging
existing ontologies. This effort is included in a project entitled “An eScience
Infrastructure for Water Quality Management in the Doce River Basin” and is concerned
with the integration of water quality data produced by various sources to assess the
impacts of the disaster that occurred in the city of Mariana, in Brazil, in 2015 (the largest
accident in history in volume of material dumped by mining tailings dams).
In this paper we show the first steps in the building of a network of interconnected
ontologies (an ontology network, as presented in (Ruy et al., 2016)) for EQR using this
approach. Section 2 presents the overview of our approach. Section 3 shows the
identification and analysis of actual environmental research data to establish the
representative concepts of the domain. Section 4 presents the identification of existing
knowledge sources and the analysis of domain coverage by them so that they can be
reused. Finally, section 5 discusses final considerations and future work.
2. The Overview of the Approach: From Data to Ontologies
Most ontology engineering methods propose an initial activity for defining the domain of
enquiry, the purpose and scope of an ontology, and relevant knowledge sources (Soares,
2009). Regarding scope definition, many methods, such as NeOn (Suárez-Figueroa et al.,
2012), opt for a top-down approach in which domain coverage arises from interactions
between ontology developers and domain experts. As we have already discussed, for the
domain of EQR, the “ontological needs” are related to the purpose of data integration.
For this case, the traditional top-down approach alone does not seem to be the most
appropriate alternative to define the ontology scope, since it does not address the origin
of the problem, that is, the EQR data. Thus, we propose a bottom-up approach for the
definition of the scope of ontologies that are aimed at data integration.
The proposed approach has as starting point the identification and analysis of
existent real-world data with the help of domain experts. By analyzing such data, it is
possible to identify the problems presented by them, but also the relevant concepts of the
domain. In other words, it provides means to define an initial scope of the ontology
network to support EQR. Once we have defined the initial scope, we need to search for
knowledge sources that address the relevant concepts of the domain so that they can be
evaluated regarding their coverage, accuracy and adequacy to the data so that they can be
reused in the ontology network. These main activities will be discussed below.
3. Environmental Quality Data Sources Identification and Analysis
In the context of our project, several sources of Brazilian water quality data were
analyzed, mainly data sources generated to assess the impacts of the Mariana disaster.
The analysis of such data highlights the problems presented by them but also helps to
identify the key concepts of the domain.
Table 1 shows water quality data from two Brazilian government agencies: data
from (IBAMA, 2018) and (IEMA, 2018) in the first two columns and from (IGAM, 2018)
in the last two. Several problems can be observed in relation to these datasets, starting
with terms used to identify them and their granularity, which vary according the source,
showing their heterogeneity. For example, (IBAMA, 2018) and (IEMA, 2018) use
“Sample Point Long Name” to identify river, state and location of collect. (IGAM, 2018)
uses “Water course” to identify river, “Counties” to identify city and state and
“Description” to identify location of sampling.
Table 1. Concepts of Water Quality Used by Different Brazilian Agencies
Another problem observed in the data of Table 1 is the measurement unit of the
water quality parameters, which is informed by (IBAMA, 2018) and (IEMA, 2018), but
not by (IGAM, 2018). In the case of data from IGAM the identification of units must be
done by whoever is interpreting the data, which may cause problems. Also, one can verify
the lack of quality of the data and vocabularies employed. For example, “Sample Type”
is used with two different purposes. When filled with Superficial, this means that the
sample material type is surface water. When it is filled with Daphnia similis, this means
that an ecotoxicological bioassay with the species Daphnia similis was performed. This
compromises clarity and can only be inferred by domain experts.
Despite the heterogeneity and the data problems of the different agencies, it is
possible to identify the relevant concepts that characterize EQR. They are (as grouped in
Table 1): responsible institute, geographic coordinates, geopolitical location, temporal
entity, sampling (sample: 62277-2016, sample material type: Superficial and sampling
procedure), quality parameter (Alkalinity of bicarbonates), measurement (measured
values, units of measurement and analytical laboratory: Merieux) and legal parameter.
4. Environmental Quality Knowledge Sources Identification and Analysis
Based on the representative concepts of EQR, a search for existing knowledge sources
(ontologies, vocabularies and standards) that cover this domain must be performed to
enable the reuse of them and avoid the creation of new resources unnecessary. Table 2
shows the relation of the resources found for the concepts identified in section 3.
Table 2. Identified Knowledge Sources
Ontology, Standard or
Description Key Concepts
Vocabulary
om-lite (Cox, 2017) Ontology for observation features, based on Feature of Interest, Observed
the O&M conceptual model from OGC and Property, Result, Procedure,
ISO 19156. Phenomenon Time and Result Time
sam-lite (Cox, 2017) Ontology for sampling features, based on the Sampled Feature, Shape, Sampling
O&M conceptual model from OGC and ISO Time, Sampling Method, Sampling
19156. Location, Current Location and Size
QUDT (Simons et al., Quantities, Units, Dimensions, Data Types Units of Measure, Quantity Kinds,
2013) ontology. Dimensions and Data Types
ChEBI (Degtyarenko et Chemical Entities of Biological Interest ChEBI Name, ChEBI ID, Definition
al., 2008) ontology. and Synonym
OP (Cox et al., 2014) Ontology for observable properties which Scaled Quantity Kind, Quality Kind,
extends QUDT, incorporating some of the Applicable Vocabulary, Property
requirements identified in the O&M model Kind, Feature of Interest, Procedure
and its successors. and Substance or Taxon
SSN (W3C SSN-XG, The Semantic Sensor Network ontology Actuatable Property, Actuator,
2005) describes sensors and their observations, the Feature of Interest, Observable
involved procedures, the studied features of Property, Result, Procedure, Sample
interest, and the observed properties. and Sensor
M-OPL (Barcellos et Measurement Ontology Pattern Language Measurable Entities, Measures,
al., 2014) addresses the measurement core Measurement Procedures,
conceptualization. Can be used for building Measurement Units & Scales and
measurement ontologies to several domains. Analysis
Darwin Core (Darwin Body of standards for biodiversity informatics. Taxon, Identification, Occurrence,
Core Task Group, 2014) It provides stable terms and vocabularies for Record level, Location, Event and
sharing biodiversity data. Material Sample
OWL-Time (Cox and Ontology that provides a vocabulary for Date Time Description, Date Time
Little, 2017) topological relations among instants and Interval, Time instant, Time interval,
intervals and information about durations and Temporal duration and Temporal
temporal position. entity
Basic Geo (WGS84 RDF vocabulary that provides a namespace for Points, Latitude, Longitude and
lat/long) Vocabulary representing lat(itude) and long(itude), using Altitude
(Brickley, 2003) WGS84 as a reference datum.
OntoBio (Albuquerque Biodiversity Ontology. Ecosystem, Environment, Spatial
et al., 2015) Location, Collect and Material Entity
These knowledge sources should be analyzed for domain coverage based on the
representative concepts of EQR. Table 3 presents the results of this analysis, showing the
concepts that are treated by each ontology, vocabulary or standard of Table 2. As can be
seen, there are already resources to deal with most of the concepts raised and they will be
analyzed for reuse. However, some of them, because they are very specific domains,
applicable only at national, state or municipal level, as in the case of legal parameters, are
not treated by existing resources and need to be structured to be coupled to the ontology
network.
The analysis of Table 3 allows us further to infer the initial modularization of the
ontology network. All the concepts of sampling are treated by four of the eleven resources
identified. In addition, the measured values and the units of measure of the measurement
group are covered, respectively, by six and four of the eleven resources identified.
Therefore, they should probably be approached as core ontologies (Ruy et al., 2016).
The other concepts are used to answer questions related to the concepts of
sampling and measurement, such as: sampling location, date and time of these activities,
measured parameters, between others. So, they should probably be approached as domain
ontologies (Ruy et al., 2016).
Table 3. Knowledge Sources and their Domain Coverage
OntoBio
sam-lite
M-OPL
Darwin
om-lite
ChEBI
QUDT
OWL-
Basic
Time
Knowledge Sources
Core
SSN
Geo
OP
Concepts
Sampling: Sample x x x x
Sample Material Type x x x x
Sampling Procedure x x x x
Measurement: Measured Values x x x x x x
Units of Measurement x x x x
Analytical Laboratory x x
Geopolitical Location x x
Geographic Coordinates x x x
Temporal Entity x x
Quality Parameters: Physical x x
Chemical x x
Biological x x x
Legal Parameters
Responsible Institution x
5. Final Considerations and Future Work
This paper presents the first steps in the building of an ontology network to support EQR.
The bottom-up approach of analyzing the real environmental quality data of several
institutions was carried out, making it possible to identify the representative concepts of
the domain, despite the heterogeneity of the data. Then, a search for knowledge sources
that cover these concepts was made. Lastly, the coverage of the concepts by the sources
was analysed, making it possible to conclude on resources that can be reused; to identify
the concepts that are not treated and, therefore, should be the subject of new ontologies
to be developed; and to establish the initial modularization of the ontology network.
As future work, we intend to follow up on the search for knowledge sources in a
systematic way so that no relevant resource is disregarded in the elaboration of the
ontology network. After, we must complete the identification of the core ontologies and
develop them. Then, we need to match the domain ontologies necessary for the modelling
of EQR and to develop those whenever their reuse is not possible. Finally, we intend to
develop an eScience portal to publicize the original EQR data annotated with
systematized metadata. The objective is for data from various sources to be integrated
and made available to the community in general.
Acknowledgments
This work is partly supported by CNPq (407235/2017-5 and 312123/2017-5), CAPES
(23038.028816/2016-41) and FAPES (69382549).
References
Albuquerque, A.C.F., Dos Santos, J.L.C., De Castro, A.N. (2015) OntoBio: A
biodiversity domain ontology for Amazonian biological collected objects, in: Proc.
Annual Hawaii International Conference on System Sciences. pp. 3770–3779.
Ashburner, M., Ball, C.A., Blake, J.A., Botstein, D., Butler, H., Cherry, J.M., Davis, A.P.,
Dolinski, K., Dwight, S.S., Eppig, J.T. (2000) Gene Ontology: tool for the unification
of biology. Nat. Genet. 25, 25.
Barcellos, M.P., Falbo, R.A., Frauches, V.G.V. (2014) Towards a measurement ontology
pattern language, in: Proc. 1st Joint Workshop ONTO.COM/ODISE, CEUR
Workshop Proceedings, Vol,. 1301.
Brickley, D. (2003). Basic Geo (WGS84 lat/long) Vocabulary. W3C Semantic Web
Interest Group, W3C.
Cox, S.J.D. (2017). Ontology for observations and sampling features, with alignments to
existing models. Semantic Web 8, 453–470.
Cox, S.J.D., Little, C. (eds.) (2017) Time Ontology in OWL. W3C Recommendation.
Cox, S.J.D., Simons, B.A., Yu, J. (2014) A Harmonized Vocabulary For Water Quality,
in: HIC2014 - 11th International Conference on Hydroinformatics. pp. 1454–1459.
Degtyarenko, K., De Matos, P., Ennis, M., Hastings, J., Zbinden, M., Mcnaught, A.,
Alcántara, R., Darsow, M., Guedj, M., Ashburner, M. (2008) ChEBI: A database and
ontology for chemical entities of biological interest. Nucleic Acids Res. 36.
FORCE11 (2016) The FAIR Data Principles [WWW Document]. La Jolla, CA
FORCE11. URL https://www.force11.org/group/fairgroup/fairprinciples
Darwin Core Task Group (2014) Darwin Core Terms: A complete historical record
[WWW Document]. TDWG. URL http://rs.tdwg.org/dwc/terms/history/
W3C SSN-XG (2005) Semantic Sensor Network Ontology [WWW Document]. online
url https//www.w3.org/2005/Incubator/ssn/ssnx/ssn.
IBAMA [WWW Document], 2018. URL http://www.ibama.gov.br/
IEMA [WWW Document], 2018. URL https://iema.es.gov.br/
IGAM [WWW Document], 2018. URL http://www.igam.mg.gov.br/
Ruy, F.B., Falbo, R.A., Barcellos, M.P., Costa, S.D. and Guizzardi G. (2016) SEON: A
software engineering ontology network, in: European Knowledge Acquisition
Workshop, pp. 527–542.
Simons, B.A., Yu, J., Cox, S.J.D. (2013) Defining a water quality vocabulary using
QUDT and ChEBI, in: Proc. 20th Intl. Congress on Modelling and Simulation.
Modelling and Simulation Society of Australia and New Zealand, pp. 2548–2554.
Soares, A. (2009) Towards ontology-driven information systems: Guidelines to the
creation of new methodologies to build ontologies, PhD dissertation, Penn. State Univ.
Suárez-Figueroa, M., Gómez-Pérez, A., Fernández-López, M. (2012) The NeOn
Methodology for Ontology Engineering, in: Suárez-Figueroa, et al. (Eds.), Ontology
Engineering in a Networked World SE - 2. Springer Berlin Heidelberg, pp. 9–34.