proposal of ontology for environmental impact assessment: an application with knowledge mobilization

Expert Systems with Applications 38 (2011) 2462–2472

Contents lists available at ScienceDirect

Expert Systems with Applications

journal homepage: www.elsevier .com/locate /eswa

Proposal of ontology for environmental impact assessment: An applicationwith knowledge mobilization

Julián Garrido ⇑, Ignacio RequenaDep. Computer Science and A.I. ETSI Informática y Telecomunicaciones, University of Granada, Daniel Saucedo Aranda s/n, 18071 Granada, Spain

a r t i c l e i n f o

Keywords:EIAEnvironmental impactOntologyKnowledge mobilizationOWL

0957-4174/$ - see front matter � 2010 Elsevier Ltd. Adoi:10.1016/j.eswa.2010.08.035

⇑ Corresponding author.E-mail addresses: jgarrido@decsai.ugr.es (J. Garrid

Requena).

a b s t r a c t

Environmental impact assessment (EIA) analyses the effects of human activity, ecosystem integrity andthe quality of the environmental services that can be provided by them. This analysis must be done priorto project execution in order to have a preventive nature. The application of new technologies in EIAneeds an adequate structure of the knowledge, so a lot of problems appear because depending of thecountry, different terminologies are used. This paper describes a proposal of ontology for EIA to establisha conceptual framework and the building process. In addition, we have developed a friendly web inter-face in order to provide easy access to the knowledge and the possibility to suggest changes in the ontol-ogy to environmental experts. The objectives are gathering and providing the EIA terminology andfacilitating structuring and the development of EIA methodologies. Finally, the paper describes howthe ontology can be used in EIA applications with knowledge mobilization.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Environmental impact assessment (EIA) is carried out usingmethodologies, which include impact identification, affected envi-ronment description, impact prediction, impact assessment and soon. The principal target in EIA consists of avoiding failures or acci-dents and environmental decay as its repair is really difficult andexpensive (Conesa, 2000) according to caution, prevention and cor-rection principles defined by European Union.

The experience and results obtained through previous projectshave enabled us to make a particular study about the diffuse rep-resentation of the most remarkable information included in theEIA, the homogenization of the different types of information in-volved in the importance impact or its magnitude, and finally,the development of software that handles fuzzy information onthe EIA infrastructure (civil engineering) (Duarte, 2000; Duarte,Delgado, & Requena, 2000, 2003, 2006; Martín, 2003) or mining(Duarte, Requena, & Rosario, 2007) and landfills (Delgado, Duarte,& Requena, 2004, 2005).

The next logical step consists of analysing the possibility ofusing similar structures to calculate the EIA in other pollutingactivities, such as the chemical industry, mining, etc. Ultimately,we intend to implement an intelligent system to assess the impactand risk caused by any activity considered in the IPPC (CouncilDirective 96/61/EC, 1996) and the directive 2004/35. Therefore,

ll rights reserved.

o), requena@decsai.ugr.es (I.

managing a common language is essential, and the representationof all the information and knowledge handled must be coherentand have the same structure.

The information handled is not always consistent; it depends onthe activities, geographic areas, governments and even countries.i.e. a name can have different meanings in some methodologiesor other activities, or the same concept can be given differentnames.

The use of ontologies enables the representation of knowledgeand the necessary processes in the EIA in various human activities,so that you can share it with other users, and in addition, the ter-minology is homogeneous. Moreover, it can be achieved by build-ing ontologies available in the network for the communityconcerned and in order to be shared and even improved by newcontributions.

This paper proposes an ontology to create a repository for gen-eral knowledge about EIA to be used by the scientific community. Itallows for the provision and gathering of the main terminology andit establishes the conceptual framework for this area to facilitatestructuring and methodology development. On the other hand,an ontology is knowledge which must be consensual. Therefore,the ontology is available in a web application,1 in order for expertsto be able to evaluate and suggest new knowledge for itsimprovement.

Depending on the context in which ontologies are used, differ-ent situations can be solved. For example, they provide a way toshare knowledge using common vocabulary, allow semantic

1 http://arai.ugr.es/eiadifusa.

J. Garrido, I. Requena / Expert Systems with Applications 38 (2011) 2462–2472 2463

tagging or knowledge exchange, provide a communication proto-col and allow the reuse of knowledge or the creation of semantic,logical and formal descriptions. Ontologies provide interoperabilityamong different systems, thanks to the qualities mentioned above.

Ontologies are currently used in different areas, such us knowl-edge engineering, artificial intelligence and computer sciences,natural language processing, knowledge representation, coopera-tive information systems, bioinformatics, e-commerce, databasedesign and integration, intelligent integration of information,information recovery and knowledge management. Moreover,ontologies can be built using different techniques like frames andfirst-order logic, descriptive logic, software engineering techniquesand database techniques (Guarino, 1997).

There are many definitions of ontologies (Gómez, Fernández, &Corcho, 2004; Gruber, 1993; Guarino, 1997); however, the follow-ing definition is clear:

‘‘A body of formally represented knowledge is based on a con-ceptualization: the objects, concepts, and other entities thatare assumed to exist in some area of interest and the relation-ships that hold among them . . . an ontology is an explicit repre-sentation of a conceptualization.” (Gómez et al., 2004)

To decide the representation of a concept, property or relation-ship in an ontology implies making a design decision. Currently,there are no standards for creating ontologies although somemethodologies suggest guidelines and rules. Some design criteriaand sets of principles are: clarity and objectivity, coherence, com-pleteness, maximum monotonic extendibility, minimal ontologicalcommitments, ontological distinction principle, diversification ofhierarchies, modularity, minimization of the semantic distance be-tween sibling concepts and standardization of names (Gómez &Benjamins, 1999; Gómez et al., 2004; Gruber, 1993).

In the practice, different problems appear during the ontologylifecycle (Kingston, 2006, 2008), such as inaccurate expert re-sponses that cause errors, the level of detail that makes reusabilitydifficult, semantic heterogeneity for the same domain, and others.

Many ontology languages have been developed during recentyears, some of them based on XML, like XOL (Ontology ExchangeLanguage), SHOE (Simple HTML Ontology Extensions) and OML(Ontology Markup Language). Afterwards, based on RDF (ResourceDescription Framework) and RDF Schema, OIL (Inference Lan-guage), DAML + OIL and lastly OWL (Ontology Web Language)were developed. OWL is a W3C (World Wide Web Consortium)recommendation and a DAML + OIL review (Gómez et al., 2004).

This paper begins with the description of the developmentmethodology that has been carried out. In the following sections(Sections 3 and 4), the most important decisions about the designand building of the ontology are explained and the most importantconcepts are described. Section 5 explains how the web applica-tion is available for the evaluation process and enables contribu-tions. Section 6 shows two examples of applications in which theontology is used to solve the EIA problem. In Section 7, the conclu-sions and remarks are shown, and finally, the references areincluded.

2. Development methodology

Frequently, each developer follows his own principles, designcriteria and phases during the development process because astandard for building ontologies does not exist although somemethodologies suggest guidelines. Several examples are the Uscholand King’s method (Uschold, 1996; Uschold & Grüninger, 1996),the method used in the KACTUS project (Schreiber, Wielinga, & Jan-sweiger, 1995), the Sensus project (Swartout, Patil, Russ, & Knight,1997), the Methontology method (Fernández, Gómez, Pazos, & Pa-

zos, 1999), etc. The disagreement about common rules and meth-ods for developing ontologies causes difficulties in differentaspects, such as the reaching of consensual ontologies for thedevelopers community, final applications, the extension of ontolo-gies by other authors and the ontologies reuse (Gómez & Benja-mins, 1999).

According to Gómez and Benjamins (1999), most of the meth-ods for building ontologies are focused on the activities develop-ment, especially on the conceptualization and ontologyimplementation, which means that not enough attention is paidto other important elements related to management, merging,learning, integration, tracing and evaluation of the ontologies.

Combining the previous methodologies explained above, wehave established a set of steps to build the proposal of EIAontology.

1. The aim and scope identification. The areas and contexts in whichthe ontology will be used may set requirements for the ontol-ogy, and they can provide an initial idea of the ontology seman-tics. Hence, the viability, the domain and objectives of theontology should be studied.

2. The ontology construction process. The ontology can be builtusing an incremental or prototype-oriented lifecycle, dependingon the necessities of the project. This step illustrates the phasesto develop the ontology.(a) Getting knowledge. It allows to identify concepts and rela-

tionships in a particular domain. The common strategiesin extracting knowledge are:

level and works down to the details to specializethem.

(ii) Bottom-up. It begins with the details and works up tothe highest conceptual level to generalize them.

(iii) Middle-out. It looks for the central concepts and gen-eralizes and specializes them.

(b) Encoding phase. The concepts, relationships and axioms areformalized in this phase using a formal language or a toolable to generate code in that language.

with a top-down strategy, an initial model will beobtained.

(ii) Improvement and structuring of the ontology is nec-essary to achieve a final design, according to modula-tion and hierarchical organization principles.

(c) Integration with other ontologies. To reuse other ontologiesto reduce development time and cost is always advisable,studying previously if the contents and the granularity aresuitable.

(d) Inference. Process of reasoning from knowledge or evidenceto obtain the final ontology.

3. Evaluation. Evaluation can be applied during the developmentprocess. This process entails an improvement process afterincluding the evaluations results. The users, experts and thetechnical evaluation allow the detection of correct and incorrectdefinitions. At this point, the consistency, completeness, redun-dancy and requirement specification must be tested.

4. Documentation. The ontology is documented according to itsaim. The main starting points, suppositions in the main con-cepts, as well as properties and primitives used in the defini-tions of concepts, must be documented.

5. Maintenance. It is especially important to establish the personresponsible for the maintenance and the way of doing it. In thisphase, the significant changes must be included in thedocumentation.

PreventiveActionasPrev

entiveAction

InustrialActivities

2464 J. Garrido, I. Requena / Expert Systems with Applications 38 (2011) 2462–2472

In the following paragraphs, the most significant points of eachstep of ontology-building are going to be briefly commented.

The aim and scope step consists of building an ontology for EIAto provide and gather the most important terminology and estab-lish the conceptual framework for this topic, to provide structuringand facilitate development of valid methodologies. The ontologyhas been developed using a prototype-oriented lifecycle in orderto have, at every moment, a prototype with a specific granularitylevel so that the experts can evaluate it.

A top-down strategy has been chosen to build the ontology sothat general concepts are firstly identified and specialized after-wards. Consequently, to obtain a prototype and better control overthe level of detail is easier.

The knowledge-extracting process has been based on Europeanlegislation whenever possible because it is a clearly accepted andrecognized international reference. Nevertheless, National legisla-tion and books related to this topic has been fallen back on whenthere was no other option. Moreover, technical reports and inter-national publications about EIA on specific occasions have beenused.

Following the W3C recommendations, OWL was used and con-sidered to implement and formalize the ontology. It has been de-signed for applications that need to process information. Theexistence of tools that reason on OWL description is one of theprincipal advantages of using OWL. These reasoners provide gener-ic support independent of the managed specific domain.

Protègè2 has been chosen for building the ontology due to itscharacteristics. Firstly, it permits the creation of ontologies inOWL, avoiding the direct treatment of OWL syntaxes. Secondly, itis an open-source platform, so it can be personalized to createknowledge models and data inputs, for example, to add vaguenessor uncertainty to the concepts (Bobillo, Delgado, & Gómez, 2006,2007). Furthermore, many plugins, which increase its functionality,have been developed for Protègè and it can work with reasoners likePellet3 or Racer.4

Other ontologies (Semantic web search) related to EIA or similartopics has been looked up. According to these studies, there are notspecific ontologies for EIA, although there are ontologies applied toother domains that include concepts related to EIA like earth sci-ences, hydrology, chemical process, ecology, pollutants, and others.Following the methodology and guidelines, to integrate theseontologies has been strongly tried. So, the integration viabilityhas been studied, analysing the contents and the granularity ofeach ontology. Some concepts have been integrated, although mostontologies were inappropriate because of unsuitable granularity orcontent.

The evaluation has been carried out during the developmentprocess. Several experts in the Environmental Technologies Areaof Civil Engineering have inspected and evaluated prototypes withdifferent granularity levels, along with a general review and contri-bution of new knowledge. The criticisms made by the experts areuseful in changing some contents and approaches (refinement).

Impact

hproduceImpact

hasIndica

dorAndM

easureUn

it

asImpactAssessment impactIn

ImpactingActions

ContaminantElement

3. Ontology design

From a technical viewpoint, EIA is an analysis process to iden-tify cause–effect relationships and to quantify, assess and preventthe environmental impact of a project (Gómez, 2003).

The main concepts of the ontology are extracted and justifiedfrom this definition. These important concepts are:

2 Developed by Stanford Medical Informatics at Stanford University School ofMedicine.

3 http://pellet.owldl.com/.4 http://www.racer-systems.com/.

� Environmental impacts.� Environmental elements that may suffer impacts.� Industrial activities.� Substances or contaminant elements.� Human actions that may produce impacts.� Environmental indicator or measure units for impacts� Impact assessment.

In the following subsections, the most important concepts andrelationships in this ontology are briefly described. These afore-mentioned concepts belong to the first hierarchical level.

It is important to remark the sources used to extract the knowl-edge. These sources are books (Barettino, Loredo, & Pendás, 2005;Block, 2000; Canter, 1995; Conesa, 2000; Garmendia, Salvador,Crespo, & Garmendia, 2005; Gómez, 2003; Seoanez, 1997; Smith,2002), technical reports (Canter & Sadler, 1997), other ontologies(GCMD ontology), standards (UNE 150008, 2000; UNE 14001,2004), PhD Thesis (Colomer, 2007; Garrido, 2008), Spanish legisla-tion (Law 11/2005, 2005; Royal Decree 1131/1988, 1988) andEuropean legislation (Council Directive 79/409/EEC, 1979; CouncilDirective 85/337/EEC, 1985; Council Directive 92/43/EEC, 1992;Council Directive 96/61/EC, 1996; Decision 2455/2001/EC, 2001;Directive 2000/60/EC, 2000). In addition, a wide range of bibliogra-phy has been used but not referenced here.

3.1. Relationships and concepts

The main concepts and relationships are shown in Fig. 1. Bothhave been extracted directly from the previous EIA definition. Onlythese concepts and relationships are represented in the figure be-cause they form the basic skeleton, although there are others in thesub-levels of the hierarchy.

These relationships contribute to improve the knowledge andthey can be used to accomplish queries from the ontology withreasoning tasks. For example, they can be used to ask for the im-pacts produced by an industrial activity or for the environmentalindicators used to assess the activity.

The concepts shown in Fig. 1 are explained in the following sub-sections. The concepts impact, impacting action and impacted ele-ment are explained in more detail than the others because they areconsidered the most important concepts in the ontology. There-fore, we have provided access to the whole ontology on the website (http://arai.ugr.es/eiadifusa).

3.1.1. ImpactThe standard UNE ISO 14001 defines environmental impact

(Block, 2000) as any adverse or advantageous change in the envi-

Indicators &MeasureUnits ImpactAssessment ImpactedElement

h

Fig. 1. Main concepts and relationships.

J. Garrido, I. Requena / Expert Systems with Applications 38 (2011) 2462–2472 2465

ronment produced by the activities, products or services of anorganization.

This concept includes the most common kinds of impacts, soimpacts are grouped depending on the environmental factors af-fected by them. These environmental factors are atmosphere, geo-physical processes, soil, habitat, landscape, socioeconomic factorsand water. Fig. 2 shows part of the impacts hierarchy and theway they has been classified.

Fig. 2. Part of the im

Concepts concerning atmosphere impacts (Barettino et al.,2005) are composition changes in the solid and gas phases, an in-crease in radioactivity, light pollution, an increase in sonorous leveland smell accumulation. Moreover, increases in fog or precipita-tion and alterations in temperatures or wind circulation are con-sidered part of the clime impact.

The geophysical impacts (Barettino et al., 2005) affect the geo-physical processes, and they are shaking, subsidence, induced se-

pact hierarchy.

2466 J. Garrido, I. Requena / Expert Systems with Applications 38 (2011) 2462–2472

ism and changes in floodable areas, waterway dynamics, erosion,sedimentation, hillside stability, surge propagation, coast flowsand aquifer recharge.

The ground impacts are classified as soil, morphology, singularelements and mineral resources. Soil is susceptible to impacts bydirect destruction, pollution or changes in its edaphic properties.The morphology impact consists of topography changes. The im-pacts in singular elements correspond with PIG (geological pointof interest) and the destruction of natural monuments. A kind ofmineral resource impact is loss of a natural resource.

The impacts on the habitat (Canter, 1995) are classified asalteration of the habitat’s properties, habitat direct loss andmovement interferences. The altered properties correspond tocomposition changes in the biotic community, a decrease in veg-etal coverage, a decrease in the critical habitat, interruption in en-ergy flow and nutrients, nutriment decrease, plague infection anda decrease in productivity. The habitat direct loss corresponds tobiodiversity reduction, critical habitat elimination, movement ofspecies with low mobility and primary production reduction.The direct loss and movement interferences correspond to move-ment of communal groups, high energy consumption, interrup-tion of the critical phases of the historical evolution, long-duration movements, migration obstruction, obstruction of accessto critical habitats and habitats of food, outside obstruction ofhostile habitats and the short-duration movement of movablespecies.

Fig. 3. Part of the impacti

The landscape impacts (Barettino et al., 2005) are visual impactand landscape quality changes.

Concepts concerning socioeconomic impact (Canter, 1995) aregrouped into price and tax changes; economic or employment ten-dencies; demographic changes; necessity of social and public ser-vices; social community; and changes in soil use, tourism andleisure.

The water impacts (Barettino et al., 2005) are grouped intogroundwater and surface water impacts. Changes in phreatic level,flow and quality are the possible impacts for groundwater. The im-pacts in the surface water are the changes in quality, radioactivitylevel, water flows and basin contribution.

3.1.2. Impacting actionsThis concept includes the actions that affect the environment;

they are the causes of the impacts. They are divided into two largegroups: the human actions (Gómez, 2003) and the natural pro-cesses, where humans are not directly responsible (GCMD ontol-ogy; Smith, 2002). Fig. 3 shows part of the impacting actionshierarchy.

There are many lists of actions in the literature because of thehigh popularity of the cause–effect matrix after 70th. The matrixincludes the actions or project activities and the environmentalelements.

The human actions are actions that affect environment, andthey are grouped into the concepts land alteration, traffic changes,

ng actions hierarchy.

5 http://www.rae.es.

J. Garrido, I. Requena / Expert Systems with Applications 38 (2011) 2462–2472 2467

production, soil transformation and construction, regimen modifi-cation, resource renovation, resource extraction, chemical treat-ment, waste treatment and waste accumulation. All of theseconcepts are detailed below.

The concept land alteration corresponds to erosion and terracecontrol, sealed mines and waste control, surface mine restoration,dredged basins and fill and marsh drainage.

The concept traffic changes corresponds to changes in the trafficof railways, trucks, cable railways, fluvial and canal transport, ves-sels, leisure sailing, pipelines, footpaths and communications.

The concept production corresponds to production of agricul-ture, cattle farming and grazing, cars and aircraft, suckle in stablesand storage of products. We have not considered here some ele-ments from the original list (Canter, 1995) to symbolize the impor-tance of these concepts. These concepts have been included in theindustrial activities, which is an entity with a higher abstractionlevel.

The concept soil transformation and construction includes thetransformation and construction of urbanizations, parcels andindustrial buildings, airports, roads, railways, elevators, bridges,electrical cabling, pipelines, corridors, barriers, dredged andaligned canals, lining canals, canals, dams, splinters, docks, ports,maritime structures, leisure places, blasting and drill, dig and fill,tunnel and subterranean installations.

The concept regimen modification corresponds to the conceptsexotic fauna, biological controls, soil coverage changes, paved andsmoothed, burning, river control and flow changes.

The concept resource renovation corresponds to reforestation,conservation and nature management, fertilizer use and wasterecycling.

The concept resource extraction corresponds to blasting anddrilling, superficial excavation, underground excavation and resto-ration, well excavation and flow extraction, clearing up and chop-ping, dredging, fishing and commercial hunting.

The concept chemical treatment corresponds to chemicaldefrosting, chemical stabilization of soil, and weed and insect con-trol with herbicides and pesticides.

The concept treatment and waste accumulation corresponds towaste accumulation, exhaust pipe and chimney emissions, liquideffluent spills, lubricants used, municipal residuals emissions, oilspills, oxidation and stabilization ponds, refrigeration water spills,scrap elimination, subterranean deposits and septic tanks.

On the other hand, natural processes are not impacting actionsby themselves; they convert into impacting actions when theyinteract with human activities. This concept is divided into theterms hydrological hazard, technological hazard, atmospheric haz-ard, geological hazard and biological hazard, all of which are de-tailed below.

Concerning atmospheric hazards, we distinguish between sin-gle and complex hazards. Single hazards are rainfall excess, ex-treme temperatures, hail, heavy snow, rays, high wind speed andfreezing rain. The complex hazards are blizzards, glaze ice, hurri-canes, stress due to heat or cold, thunderstorms and tornadoes.

The biological hazards correspond to animal and plant invasion,epidemics and forest or grassland fires.

The geological hazards are grouped into earthquakes, massmovement, volcanic eruptions, rapid sediment movement, sedi-mentation and erosion.

The hydrological hazards correspond to droughts, floods,freezes, groundwater flow discharge, infiltration, land subsidence,percolation, runoff, salt water intrusion and thaws.

The technological hazards correspond to accidental release oftoxic substances; biological, chemical or nuclear warfare; collapseof public buildings or other, bigger structures; explosion andindustrial fires; nuclear power plant failures; and accidents intransport.

3.1.3. Industrial activitiesAccording to the Royal Academy of the Spanish Language,5

industry is the set of procedures accomplished to obtain, transportor transform one or several natural products. Industrial activitiesform a sector with its own troubles from an environmental view-point and with a great importance in EIA.

Experts insisted on using a classification of industry activitiesbased on European directives like the directive 96/61/CE (IPPC),concerning integrated pollution prevention and control, in whichinstallations or parts of installations used for research, develop-ment and the testing of new products and processes are notincluded.

According to Council Directive 96/61/EC (1996), industrialactivities are classified as chemical industry, energy industry, pro-duction and metal transformation, waste management and otherindustries that cannot be included with the previous concepts.These include carbon or electrographite production, milk treat-ment and processing, paper and paperboard production, slaughter-houses, poultry or pig intensive rearing and treatment andprocessing of food products.

3.1.4. Impacted elementThe elements or environmental factors are affected by the im-

pacts. There are many examples of classifications for environmen-tal factors, such as the European Union (EU) classification in thedirective 85/337 on the assessment of the effects of certain publicand private projects on the environment. This directive establishesthat EIA will identify, describe and assess the direct or indirect ef-fects on humans, fauna, flora, soil, water, climate and air, interac-tion between the previous factors, material assets and culturalheritage.

There are not major differences between the classifications fol-lowed by different authors; therefore, we have not used a concreteclassification but have collected the concepts from several classifi-cations to obtain a complete and structured one.

All the environmental factors are grouped into categories suchas land surface, landscape, process, living things, water, habitat,atmosphere and socioeconomic elements.

The hierarchy of the water concept is shown in Fig. 4 as thewater classifications, especially the surface water which is dis-cussed with profusion in the literature.

3.1.5. Preventive actionThis concept corresponds to actions used to prevent or hinder

environmental damages, and to decrease damage probability andthe severity of its consequences. Some examples are security sys-tems, redundancy, preventive maintenance activities and so on(UNE 150008, 2000).

3.1.6. Indicators and measure unitsThis concept corresponds to a simple measure of environmental

factors or biological species, assuming that these measures are rep-resentative in biological or socioeconomic systems (Canter, 1995).

An environmental indicator (Garmendia et al., 2005) is an ele-ment that contributes information about the state of the ecosystemor something relative to it. Furthermore, biological indicators areused where the existence of a plant or animal in an area is stronglyindicative of specific environmental conditions. For example, ex-perts select the indicators depending on their sensibility to or tol-erance of pollution or its consequences (Canter, 1995).

In addition, an indicator is the expression used to calculate it.Sometimes, the indicator is measured in an indirect way with amodel (Conesa, 2000).

Fig. 4. Water hierarchy.

2468 J. Garrido, I. Requena / Expert Systems with Applications 38 (2011) 2462–2472

Indicators can be classified from a practical view point (Gar-mendia et al., 2005) on alarm, alert, sensibility and integrationindicators.

3.1.7. Impact assessmentImpact assessment is a part of the EIA process and it is included

in the technical report when the environmental impact study isperformed. The environmental assessments listed in Royal Decree1131/1988 (1988) has been collected.

3.1.8. Contaminant elementAccording to Council Directive 96/61/EC (1996) and Directive

2000/60/EC (2000), the contaminant elements are substances pro-duced as a result of human activity; therefore, they may be harmfulto human health or environmental quality, damage material assets,or impair or interfere with amenities and other legitimate uses ofthe environment.

This concept includes substances that are considered pollutants,such as acid waters in mining, artificial radionuclides, chemical re-agents, leachates, tailings, and according to European legislation,air pollutants (Council Directive 96/61/EC, 1996) and prioritysubstances for water (Directive 2000/60/EC, 2000).

3.2. Properties

The class Properties groups concepts that are not very signifi-cant in the EIA process but that are useful for describing others.These concepts are particular characteristics related to the rest ofthe concepts using hasProperty connections. Moreover, they areused in formal definitions.

These properties have been collected from the same sourcesused in the previous concept hierarchies.

3.3. Other important concepts

This section lists other important concepts included in theontology such as methodology, environmental hazard, scene,

development risk, mechanism for repairing environmental dam-ages, environmental risk assessment, environmental impactassessment, vigilance and control schedule.

These concepts have not been extracted from the EIA definitionalthough they are important in EIA, in contrast with the conceptsdescribed in previous sections.

4. Evaluation, documentation and maintenance

The evaluation phase can proceed in a strict order, without anyoverlapping steps in the methodology. It involves a serious disad-vantage and it is difficult to go back and change it if something wasnot well-thought; therefore, a continuous revision process ispreferred.

The work of environmental experts has been important in guid-ing the development of the ontology. They have provided thesources in order to study this issue. Furthermore, they have identi-fied deficiencies, and their criticisms have been added to theontology.

Moreover, a web interface has been developed to make the eval-uation and maintenance processes easier. Consequently, expertscan inspect the ontology and suggest proposals to improve it andreach a consensus. Currently, the web application is running andexperts have been requested to contribute.

The documentation has been parallelized with the developmentprocess. The documentation contains the main starting points, sup-positions, conjectures or hypotheses. It also describes every meth-odology step and justifies all design decisions and concepts orrelationships.

5. Web application for consulting and contributions

A web application has been developed for easy consultation ofthe ontology without installing any applications, and it is espe-cially focused on environmental experts who are not familiar withOWL and similar technologies. Furthermore, the web application

J. Garrido, I. Requena / Expert Systems with Applications 38 (2011) 2462–2472 2469

facilitates jointly consulting and making contributions. Anotherreason for this application is that this ontology has a high levelof requirements in expert evaluation in order to be a referencefor the development of methodologies.

The web interface based on PHP provides experts access toinformation and work without the complex syntax of OWL.

The interface shows a window divided into two parts. The firstone corresponds to the hierarchy of concepts; it can be exploredexpanding and contracting the concepts, like folders in WindowsExplorer. Moreover, important details (properties and definitions)for experts are shown in the second part of the window when anode is selected.

The purposes of the web application are evaluation and knowl-edge homogenization, so experts can easily contribute criticisms orsuggestions. Several buttons appear in the interface when you se-lect a concept, and they allow you to make contributions for the se-lected concept or a general contribution to the ontology by fillingout a simple form. What’s more, they allow you to consult the con-tributions of other experts.

The web application has been built according to Fig. 5 and byusing our own experience in the development of a Final Applica-tion Generator based on model-driven software development(Garrido, Martos, & Berzal, 2007).

First of all, the generation process begins by pre-processing theontology in OWL. Then, an intermediate representation is pro-duced in XML. And finally, all the files required by the PHP applica-tion are generated from the intermediate representation by a XSLTtransformation.

The routine work needed to create a working application isautomated, thus avoiding the normal mistakes in a manual pro-cess; it would entail and improve the overall application quality.Therefore, to change the ontology, we only have to change the in-volved concepts and re-generate completely the web application inone step.

Once the application has been generated, we can use the webinterface normally and the contributions will be stored dynami-cally in a database.

6. EIA ontology applications

Ontologies are used in many problems and contexts; it is a kindof solution habitual in the scientific community. In this paper twoexamples of how the proposal EIA ontology can be used, areshown. The first one consists of an EIA system based on brief ontol-ogy architecture (Delgado, Pérez-Pérez, & Requena, 2005). The sec-ond example consists of a knowledge mobilization architecturebased on Context-Domain Significance models (CDS) (Gómez,2008).

6.1. EIA with brief ontologies

Brief or re-addressable ontologies are defined as the ontologythat includes a small amount of knowledge referring to conceptsexisting in an ontology more generic. They are proposed for dealing

Fig. 5. Generation process of the web application.

with the problems encountered in knowledge mobilization fromlow performance devices like PDAs, mobile phones and so on.

The system requires the existence of a generic ontology, whichmeans a well structured and defined knowledge base for storinginformation. The brief ontologies behave like indirect addressingwhere data are not accessed directly and it is provided an addresswhere the data is. The addressing allows to the user filter the infor-mation before retrieving it.

The re-addressable ontology gives to the user pairs of conceptand value (keywords) which are consulted from the brief ontology.

They can be used if the user does not want to work with a heavyontology due to efficiency reasons or he is only interested in usinga reduced portion. Using the brief ontology, better system behav-iour can be obtained because of users retrieve filtered informationand they only obtain what they need. Hence, transferring uncon-nected data that exists in the whole ontology is not necessary.

Therefore, the brief ontologies are useful if the system is work-ing with bandwidth limitations to avoid collapsed communicationsand its loss of effectiveness. In addition, they are useful in applica-tions with reduced capability of storage or computation. These lim-itations are characteristics of the mobile architectures so bettersystem response time and more efficient use of the device can beachieved with brief ontologies.

The information and data needed in EIA are different dependingon the problem or the evaluated activity. In the EIA applicationcontext, the brief ontology indicates the information and datawhich need to be gathered to carry out the EIA process, so thatthere will be a brief ontology for each different EIA process. Forexample, the required information in EIA for desalinization plant,mining or a ferrous metal foundry is very different, although thereare obviously some common elements. In fact, the elements takenin account for the same activity in the EIA process can be some-times different depending on the country due to its ownlegislation.

According to the definition and Fig. 6, there is a generic ontol-ogy and a brief ontology for each particular situation. An environ-mental study of a cheese factory is tackled in order to show thesystem overview. In Kotoupas, Rigas, and Chalaris (2007), a Ched-dar cheese factory assesses five alternative scenarios for the treat-ment of cheese whey wastewater. In this case they only need a setof concentrations and daily throughputs of the environmental andaqueous properties of the stream (see Table 1). This is a simpleexample of EIA due to it only uses several standard indicators,although many others complex examples can be used, such asEIA methodologies described in Conesa (2000) and Canter (1995).

The system is thought with a distributed architecture in whichthe mobile devices request information contained in data sources.A multi-agent system will be responsible for doing the necessarytasks in order to consult the brief ontologies, which will be read-dressed to the requesting devices.

Using the device, the environmental expert will request thenecessary information for the EIA process in a cheese factory (re-addressable ontology). After that, the agents will gather the dataand they will respond by sending the re-addressable ontology withthe knowledge connected with the request in order to be able to dothe EIA. However, the user does not deal directly with the agents.There is a middle layer to automate the interaction between userand system based on web services.

The brief ontology used in the Cheddar cheese factory evalua-tion contains two main entities. The first one is a set of the stan-dard indicators needed to accomplish the EIA process. This set ofenvironmental indicators is composed of a selection from the en-tire list in the ontology because they are considered relevant inthe treatment and processing of milk. Moreover the brief ontologycontains additional information like the minimum and maximumvalues allowed.

Fig. 6. Brief ontologies schema.

Table 1Acronyms for standard indicators used in Kotoupas et al. (2007).

TOC Total organic carbonCOD Chemical oxygen demandThOD Theoretical oxygen demandBODu Biochemical oxygen demandBOD5 Biochemical oxygen demandTKN Total Kjeldahl nitrogenNH3 Ammonia nitrogenNO3/NO2 Nitrate/nitrite nitrogenTP Total phosphorousTS Total solidsTSS Total suspended solidsVSS Volatile suspended solidsDVSS Degradable volatile suspended solidsTDS Total dissolved solidsVDS Volatile dissolved solidsDVDS Degradable volatile dissolved solidsCaCO3 Calcium carbonate

2470 J. Garrido, I. Requena / Expert Systems with Applications 38 (2011) 2462–2472

The second entity contained in this brief ontology is the hierar-chy of environmental factors which can be affected by this indus-

try. Furthermore, the relationship between the indicators and theenvironmental factors affected by them is added as well. Thisknowledge will allow to the environment expert obtain not onlyan index of the Cheddar factory but also an index of every environ-mental factor.

Hence, the environmental experts only manage the essentialknowledge, in this case, environmental factors and indicators;and they avoid dealing with needless knowledge in the momentthey are taking the measures for the indicators. Consequently otherrelated knowledge in the generic ontology that is not taken in ac-count now, will be used in later EIA phases. For example, a deeperassessment would consider the environmental impacts of theactivity or the preventive actions. So that other brief ontologiesare defined for the other phases of the EIA process.

To sum up, these data are necessaries to carry out the EIAaccordingly with this method, so the information is contained inthe brief ontology, thus dealing with the whole ontology and irrel-evant information are avoided by mobile devices.

6.2. EIA with the Context-Domain Significance model

Nowadays, information and knowledge based systems needprocess a great deal of information and manage the problem ofthe information overload. Many approaches have been developedto deal with this topic.

The problem of emergency-assistance is presented in Gómez(2008). The authors have developed a system which extracts therelevant information from his clinical history because the clinicalstaff cannot spend such a long time reading a large clinical historyso that they need a summary with the most important information,like the adverse drug events.

They have designed an architecture and a model for knowledgemobilization to tackle the problem. The architecture is based on aContext-Domain Significance (CDS) model to establish a patternover an ontological knowledge base for representing significanceof the context-depending information.

This proposal uses the CDS model to extract the relevant knowl-edge from a database where data is stored due to they are dealingwith a matter of urgency. By contrast, EIA does not have the ex-treme necessity of summarizing the data for the survival of some-one. However, it leads us to improve its knowledge management.

Supposing an employee who needs to accomplish a lot of envi-ronmental impact assessments for a wide range of human activi-ties, many variables are needed by each activity and most ofthem are different depending on the kind of evaluated activity.Consequently, a CDS model can be used in an EIA context.

As opposed to Gómez (2008), this approach would be focusedon distinguishing which is the relevant knowledge that has to begathered by the employee.

The CDS model requires three different ontologies: the contextmodel, the domain model and the CDS model. Although contextand domain model could be defined in different ontologies or inan only monolithic ontology, the CDS model has to be built sepa-rately. These are briefly described below.

The domain ontology (OD) contains the specific knowledgewhich is needed to solve the analysed problem, a particular areaof activity or interest. Moreover, the concepts of OD can be usedto defined more complex concepts expressions and enrich thedomain.

Context is the set of facts or circumstances that surround a sit-uation or event. The context ontology (OC) allows to define the sce-nario for a particular problem. As in the domain, complex contextexpressions can be defined by using the OC elements.

The target of the CDS ontology is to connect the domain expres-sions and context descriptions so that it formally defines which the

Table 2Example of a CDS model considering the elements of Table 1.

Profile 1. Milk treatmentC1 �MilkTreatmentAndProcessingD1 � TOC \ COD \ ThOD \ BODu \ TKNP1.1 � $hasAction � C1 \ $hasIndicator � D1

Profile 2. Milk treatment, rise and care of animals in a stableC2 �MilkTreatmentAndProcessing\RiseAndCareAnimalsInStables

D2 � TOC \ COD \ ThOD \ BODu \ TKN\AffectedAreaBySmells\ScentedMaterialsConcentration

P2.2 � $hasAction � C2 \ $hasIndicator � D2

J. Garrido, I. Requena / Expert Systems with Applications 38 (2011) 2462–2472 2471

information of interest is by using profiles which are defined withquantified roles.

On the one hand, the domain expressions represent the knowl-edge which has to be filtered. On the other hand, the contextdescriptions represent a particular situation among the differentpossibilities, whereas the CDS ontology shows the relevance ofthe domain in a specific context. Generally speaking, the systemhas to show the relevant or significant knowledge (domain) in aparticular context or situation using the CDS ontology.

An straightforward example of a CDS model with two differentprofiles for the Cheddar cheese factory is shown in Table 2. Theo-retically, both of them should contain the whole list of environ-mental indicators belonging to the Table 1 although it has notincluded in order to make the example easier and moreunderstandable.

The first case establishes the environmental indicators (domain)for a factory whose main function is the milk treatment and pro-cessing (context), it is a kind of industrial activity. The profile P1.1

establishes the relationship between the complex context C1 andthe complex domain D1 by mean of the properties hasAction andhasIndicator.

The second one shows how the environmental impact assess-ment for complex human activities can be done. Supposing a gen-eral activity which is not included in the OC ontology; the CDSmodel should include a complex context expression describing thisactivity and using the concepts of the OC ontology.

The second example corresponds with a cheese factory whichraises their animals in stables to obtain milk. Consequently, thecontext would be the combination of two activities which existin the OC ontology, an activity for milk treatment and processingand an activity for rising and caring animals in stables.

As we can see, the EIA ontology can be used to define the con-text and the domain expressions for a final application and finally,the CDS ontology only has to establish which environmental indi-cators correspond with the relevant domain and which actions orindustrial activities correspond with the context.

7. Conclusions

An ontology for EIA has been presented. The ontology has beenbuilt because previous works have encouraged us to use moresophisticated techniques in knowledge management due to thelarge quantity of variables and the complexity of the different pro-cesses involved in EIA.

The ontology was built with two basic purposes. The first oneconsist in using the ontology directly like a reference for develop-ers of EIA methodologies in order to give access to structured infor-mation so that natural language definitions are included.

This purpose has required to build the ontology rigorously andstrictly, taking in account the most specialized literature and mak-ing an arduous and demanding development. So, large documenta-

tion has been written because every concept or group of conceptshas been described and justified.

The second purpose is using the ontology to manage the knowl-edge in an expert system for EIA so that formal definitions are in-cluded for reasoning tasks.

Subsequently, a web application has been built for consultingthe ontology avoiding complex interfaces. The tool is also thoughtto validate the contents requesting any suggestion or constructivecriticism of the ontology by mean of web forms in order to achieveas consensual knowledge as possible.

Besides, two applications using this ontology are presented.Firstly, an application based on brief or re-addressable ontologiesusing this proposal as a generic ontology together with small partsof it as brief ontologies in order to consult filtered information.

The second application is based on a CDS model where the con-text and domain ontology is built from the EIA ontology. This appli-cation allows to extract relevant knowledge which is considereduseful in different situations.

Acknowledgements

This work has been partly supported by projects of ‘‘Junta deAndalucía”: (CICE) P07-TIC-02913 and P08-RNM-03584.

References

Barettino, D., Loredo, J., & Pendás, F. (Eds.). (2005). Soil and water acidification:Problems and solutions. Geological and Mining Institute of Spain (in Spanish).

Block, M. (2000). Identifying environmental aspects and impacts. ASQ Quality Press.Bobillo, F., Delgado, M., & Gómez, J. (2006). A crisp representation for fuzzy SHOIN

with fuzzy nominals and general concept inclusions. In Proceedings of the 2ndISWC workshop on uncertainty reasoning for the semantic web (Vol. 218).

Bobillo, F., Delgado, M., & Gómez, J. (2007). Optimizing the crisp representation ofthe fuzzy description logic SROIQ. In Proceedings of the 3rd ISWC workshop onuncertainty reasoning for the semantic web.

Canter, L. (1995). Environmental impact assessment. McGraw-Hill.Canter, L., & Sadler, B. (1997). A tool kit for effective EIA practice: Review of methods

and perspectives on their application. A supplementary report of the internationalstudy of the effectiveness of environmental assessment. Tech. rep.,Environmental Ground Water Institute University of Oklahoma, Institute ofEnvironmental Assessment, UK, International Association for ImpactAssessment, USA.

Colomer, F. (2007). Analysis and systematization of environmental security in urbanlandfills. Application to lixiviate ponds. Ph.D. thesis, Valencia University (inSpanish).

Conesa, V. (2000). Methodological guide for environmental impact assessment. Mundi-Prensa (in Spanish).

Council Directive 79/409/EEC (1979). Council Directive 79/409/EEC of 2 April 1979on the conservation of wild birds (OJ L 103, 25.4.1979, p. 118).

Council Directive 85/337/EEC (1985). Council Directive 85/337/EEC of 27 June 1985on the assessment of the effects of certain public and private projects on theenvironment (OJ L 175, 5.7.1985, p. 4048).

Council Directive 92/43/EEC (1992). Council Directive 92/43/EEC of 21 May 1992 onthe conservation of natural habitats and of wild fauna and flora (OJ L 206,22.7.1992, p. 750).

Council Directive 96/61/EC (1996). Council Directive 96/61/EC of 24 September1996 concerning integrated pollution prevention and control (OJ L 257,10.10.1996, p. 2640).

Decision 2455/2001/EC (2001). Decision 2455/2001/EC of the European Parliamentand of the Council of 20 November 2001 establishing the list of prioritysubstances in the field of water policy and amending Directive 2000/60/EC (OJ L331, 15.12.2001, p. 15).

Delgado, M., Duarte, O., & Requena, I. (2004). Evaluation of investment alternativesusing financial and non-financial information. In EUROFUSE’04, Poland.

Delgado, M., Duarte, O., & Requena, I. (2005). A fuzzy methodology for theenvironmental impact evaluation of garbage dumps. In Eleventh internationalfuzzy systems association world congress (IFSA’05).

Delgado, M., Pérez-Pérez, R., & Requena, I. (2005). Knowledge mobilization throughre-addressable ontologies. In Proceedings of the EUSFLAT05 (pp. 154–159).

Directive 2000/60/EC (2000). Directive 2000/60/EC of the European Parliament andof the Council of 23 October 2000 establishing a framework for communityaction in the field of water policy (OJ L 327, 22.12.2000, p. 173).

Directive 2004/35/CE (2004). Directive 2004/35/CE of the European Parliament andof the Council of 21 April 2004 on environmental liability with regard to theprevention and remedying of environmental damage (OJ L 143, 30.4.2004, p.5675).

Duarte, O. (2000). Application of fuzzy techniques in environmental impact assessment.Ph.D. thesis, Granada University (in Spanish).

2472 J. Garrido, I. Requena / Expert Systems with Applications 38 (2011) 2462–2472

Duarte, O., Delgado, M., & Requena, I. (2003). Algorithms to extend crisp functionsand their inverse functions to fuzzy numbers. International Journal of IntelligentSystems, 18, 855–876.

Duarte, O., Delgado, M., & Requena, I. (2006). An arithmetic approach for thecomputing with words paradigm. International Journal of Intelligent Systems, 21,121–142.

Duarte, O., Delgado, M., & Requena, I. (2000). Application of the extension of crispfunctions to fuzzy numbers in the environmental impact analysis. InProceedings of the IPMU2000 (pp. 134–140).

Duarte, O., Requena, I., & Rosario, Y. (2007). Fuzzy techniques for environmentalimpact assessment in the mineral deposit of Punta Gorda (Moa, Cuba).Environmental Technology, 28, 659–669.

Fernández, M., Gómez, A., Pazos, J., & Pazos, A. (1999). Ontology of tasks andmethods. IEEE Intelligent Systems and Their Applications, 14(1), 37–46.

Garmendia, A., Salvador, A., Crespo, C., & Garmendia, L. (2005). Environmental impactassessment. Pearson Educación (in Spanish).

Garrido, M. (2008). Environmental diagnostic methodology in landfills, adaptation touse with fuzzy techniques and to apply in Andalucía landfills. Ph.D. thesis, GranadaUniversity (in Spanish).

Garrido, J., Martos, M. A., & Berzal, F. (2007). Model-driven development usingstandard tools. In Proceedings of the 9th international conference on enterpriseinformation systems (ICEIS 2007) (pp. 100–103).

GCMD ontology. Semantic web for earth and environmental terminology. <http://sweet.jpl.nasa.gov/ontology>.

Gómez, D. (2003). Environmental impact assessment: A preventive tool forenvironmental management. (Mundi-Prensa). in Spanish.

Gómez, J. (2008). Knowledge mobilization: Architectures, models and applications.Ph.D. thesis, Granada University.

Gómez, A., & Benjamins, V. (1999). Overview of knowledge sharing and reusecomponents: Ontologies and problem-solving methods. In Proceedings of theNetherlands/Belgium conference on artificial intelligence (BNAIC) (p. 242).

Gómez, A., Fernández, M., & Corcho, O. (2004). Ontological engineering, with examplesfrom the areas of knowledge management, e-commerce and the semantic web.Springer-Verlag London Ltd.

Gruber, T. (1993). In N. Guarino & R. Poli (Eds.), Formal ontology in conceptualanalysis and knowledge representation. Deventer, The Netherlands: KluwerAcademic Publishers.

Guarino, N. (1997). Understanding, building and using ontologies. InternationalJournal of Human–Computer Studies, 46, 293–310.

Kingston, J. (2006). Multi-perspective modeling: A framework for knowledgerepresentation and knowledge management. Tech. rep. 45, Joseph Bell Centre,School of Law, University of Edinburgh.

Kingston, J. (2008). Multi-perspective ontologies: Resolving common ontologydevelopment problems. Expert Systems with Applications, 34, 541–550.

Kotoupas, A., Rigas, F., & Chalaris, M. (2007). Computer-aided process design,economic evaluation and environmental impact assessment for treatment ofcheese whey wastewater. Desalination, 213, 238–252.

Law 11/2005 (2005). Law 11/2005 of 22 June 2005 amending the law 10/2001 of 5July of the National Hydrological Plan (BOE num. 149, of 23 June 2005) (inSpanish).

Martín, J. M. (2003). Fuzzy multicriteria patterns: Applications. Ph.D. thesis, GranadaUniversity (in Spanish).

Royal Decree 1131/1988 (1988). Establishing the regulations of the perform of theRoyal Decree 1302/1986 of 28 June, of environmental impact assessment (BOEnúm. 239, of 5 October 1988) (in Spanish).

Schreiber, G., Wielinga, B., & Jansweiger, W. (1995). The KACTUS view on the Oword. In IJCAI workshop on basic ontological issues in knowledge sharing.

Semantic web search. <http://swoogle.umbc.edu>.Seoanez, M. (1997). Environmental engineering applied: Practical cases. Mundi-Prensa

(in Spanish).Smith, K. (2002). Environmental hazards, assessing risk and reducing disaster.

Routledge.Swartout, W., Patil, R., Russ, T., & Knight, K. (1997). Toward distributed use of large-

scale ontologies. In Spring Symposium Series on Ontological Engineering(pp. 33–44). Stanford: AAAI Press.

Treaty of Amsterdam Amending the Treaty on European Union (1997). The treatiesestablishing the European communities and certain related acts (OJ c 340,10.11.1997, p. 93).

UNE 14001 (2004). Environmental management system. Requirements for its use.UNE 150008 (2000). Environmental risk analysis and evaluation.Uschold, M. (1996). Building ontologies: Towards a unified methodology. Expert

Systems, 96.Uschold, M., & Grüninger, M. (1996). Ontologies: Principles, methods, and

applications. Knowledge Engineering Review, 11(2), 93–155.