linked data for potential algal biomass production monika solanki / linked data for potential algal...

Undefined 1 (2009) 1–5 1IOS Press

Linked data for Potential Algal BiomassProductionEditor(s): Krzysztof Janowicz, University of California, Santa Barbara, USASolicited review(s): Anne Thessen, Arizona State University, USA; Boris Villazon-Terrazas, iSOCO, Madrid, Spain; Femke Reitsma,University of Canterbury, New Zealand

Monika Solanki a,∗, Johannes Skarka b, Craig Chapman c

a Knowledge Based Engineering Lab, Birmingham City University, Birmingham, United KingdomE-mail: [email protected] Karlsruhe Institute of Technology, ITAS, Karlsruhe, GermanyE-mail: [email protected] Knowledge Based Engineering Lab, Birmingham City University, Birmingham, United KingdomE-mail: [email protected]

Abstract. In this paper we present an account of the publication of a suite of datasets, LEAPS, that collectively enable theevaluation of potential algal biomass production sites in North West Europe (NWE). LEAPS forms the basis of a prototypeWeb application that enables stakeholders in the algal biomass domain to interactively explore via various facets, potential algalproduction sites and sources of their consumables across NWE.

Keywords: Algae, Biomass, Energy, triplification, linked data, Ontologies

1. Introduction

Recently algal biomass has been identified as a po-tential source of large scale production of biofuels.In order to derive fuels from biomass, algal operationplant sites are setup that facilitate biomass cultivationand conversion of the biomass into end use products,some of which are biofuels. In this paper we presentLEAPS - Linked Entities for Algal Plant Sites, a suiteof linked datasets that collectively enable the evalua-tion of the potential of algal biomass production sitesin North West Europe (NWE). The framework under-lying LEAPS has been developed within the context ofthe EnAlgae1 project. In Section 2 we present the mo-tivation behind curating the LEAPS dataset suite. Sec-tion 3 provides an account of the raw datasets which

*Principal and Corresponding author. Email:[email protected]

1http://www.enalgae.eu/

served as the basis for LEAPS. Section 4 illustrates thevocabularies used in the datasets. Section 5 discussescharacteristics of the dataset. Section 6 describes theprototype Web application built over LEAPS2. Section7 outlines limitations and finally Section 8 presentsconclusions and discusses future work.

2. Motivation

The idea that algae biomass based biofuels couldserve as an alternative to fossil fuels has been em-braced by councils across the globe. Major companies[5,6], government bodies [8] and dedicated non-profitorganisations such as ABO (Algal Biomass Organisa-tion) 3 and EABA (European Algal Biomass Associ-

2http://www.semanticwebservices.org/enalgae/

3http://www.algalbiomass.org/

0000-0000/09/$00.00 c© 2009 – IOS Press and the authors. All rights reserved

2 Monika Solanki / Linked data for Potential Algal Biomass Production

ation)4 have been pushing the case for research intoclean energy sources including algae biomass basedbiofuels.

Within the context of algae production, a major ob-jective of the EnAlgae project is to create a network ofpilot scale algal facilities across NWE in order to ad-dress the current lack of verifiable information on algalproductivity. The integrated network incorporates anup to date inventory in which pilots collect and sharedata in a standardised manner and provide demonstra-tions to diverse project partners, observers and stake-holders.

One of the key gaps identified within the algalbiomass domain is the lack of a semantically en-riched infrastructure for sharing and reusing knowl-edge. An introspection of the algae-to-biofuels lifecy-cle reveals several layers where Semantic Web stan-dards and linked data technologies could be very suc-cessfully applied and immensely benefit the commu-nity. Algal biomass data manifests itself across severalfacets. At a very high level, the value chain for algalbiomass ranges from cultivation of algae to produc-tion of biofuels and other products from the cultivatedbiomass [4]. Figure 15 depicts a schematic represen-

Fig. 1. The Algal biomass supply chain

tation of the algal biofuel value chain stages and thecontributions that Semantic Web and linked data couldbring to each of the stages.

Stage 1 encompasses the cultivation of algae. It in-volves setting up an algal cultivation site and incorpo-rates datasets about location related geographical in-formation about the sites, locations of sources of light,CO2, nutrients, water and labour. The linked data for

4http://www.eaba-association.eu/5All figures in the paper are available at http://purl.org/

biomass/LEAPSFigures

datasets is described using domain specific and spa-tial ontologies. Stage 2 is concerned with the harvest-ing of algal biomass. Datasets and vocabularies relatedto harvesting strategies and extraction techniques arethe key semantic outputs of this stage. Stage 3 involvesthe conversion of biomass to end use products such asbiofuels and other constituents. Application ontologiesand product ontologies such as GoodRelations6 will becrucial in describing the datasets for this stage.

In this paper we showcase the publication of linkeddata for some of the datasets from stage 1. The ob-jective of LEAPS is to enable the stakeholders ofthe algal biomass domain to interactively explore,via linked data, potential algal production sites andsources of their consumables across NUTS (Nomen-clature of Units for Territorial Statistics)7 regions inNWE.

3. Transformation of Raw datasets to linked data

Table 1 highlights the sources of the datasets alongwith their purpose. All the datasets were openly avail-able in non-RDF formats with various origins. By per-forming potential analysis on different NUTS levels,regions with high potential can be identified. The cal-culations are based on high resolution (300 m) dataon possible algae production sites and data on CO2sources.

The transformation of the raw datasets to linked datatakes place in two steps. The first part of the data pro-cessing and the potential calculation are performed ina GIS-based model which was developed for this pur-pose using ArcGIS 8 9.3.1. Raw datasets with variousorigins and formats are first transformed using bespokecomputational algorithms to an ArcGIS specific XMLformat. This step is very crucial for two main reasons:

– It brings uniformity in the format of representa-tion of the datasets.

– In the process of transformation, important com-putations that are part of the final datasets are per-formed.

The second step of lifting the data from XML to RDFis carried out using a bespoke parser that exploits

6http://purl.org/goodrelations/v17http://bit.ly/I7y5st8http://www.esri.com/software/arcgis/

index.html

Monika Solanki / Linked data for Potential Algal Biomass Production 3

XPath 9 to selectively query the XML datasets10 andgenerate linked data using the ontologies illustrated inFigure 3 and a linking engine. While in most cases,transforming XML datasets to their linked data coun-terparts is done assuming a simplistic one-to-one map-ping between the XML elements and RDF entities, inour scenario, the original data sources had several lim-itations and a one-to-one transformation was not possi-ble. A bespoke engine [7] was developed that enabledthe transformation for each of the datasets.

An architecture underlying the transformation pro-cess is depicted in 2. The main components of the ap-

Fig. 2. Architecture of LEAPS

plication are

– Parsing modules: As shown in Figure 2, the pars-ing modules are responsible for lifting the datafrom their original formats to RDF. The liftingprocess takes place in two stages to ensure unifor-mity in transformation.

– Linking engine: The linking engine along withthe bespoke XML parser is responsible for pro-ducing the linked data representation of thedatasets. The linking engine uses ontologies,dataset specific rules and heuristics to generateinterlinking between the five datasets. From the

9http://www.w3.org/TR/xpath/10A snippet of one of the XML dataset is available at

http://www.purl.org/biomass/XMLSnippet

LOD cloud, we currently provide outgoing linksto DBpedia11 and Geonames12.

– Triple store: The linked datasets are stored in atriple store. We use OWLIM SE 5.0 13.

– Web services: Several REST Web services havebeen implemented to provide access to the linkeddatasets.

– SPARQL endpoints: SPARQL endpoints thatprovide access to individual dataset repositoriesare available. Snorql has been customised as thefront end for the endpoint. An endpoint for feder-ated queries is planned to be implemented as partof future work.

– Ontologies: A suite of OWL ontologies for thealgal biomass domain have been designed andmade available.

– Interfaces: The Web interface provides an in-teractive way to explore various facets of sites,sources, pipelines, regions, ontolgoies and SPARQLendpoints. The map visualisation has been ren-dered using Google maps. Besides the SPARQLendpoint and the interactive Web interface, aREST client has been implemented for access tothe datasets. Query results are available in RD-F/XML, JSON, Turtle and XML formats.

– Biological taxonomy visualisation: A subset ofthe Algaebase database which is the largest infor-mation source of algae on the Web, has been re-trieved and curated in our triple store. This datasetwhen integrated with the dataset for algal cultiva-tion site, can inform stakeholders about the strainsof algae that can be harvested on that site. Further,the Semantic Import plugin14 of Gephi15 has beenexploited to visualise the biological taxonomy ofalgae. This visualisation is also made availablevia the LEAPS interface.

For each of the algae production site, information onbiomass yield and site area are determined. Addition-ally, data on CO2-providing industrial or power plantscan be queried for each site and costs for CO2-supplycan be calculated. Thus, the data enables a screen-ing for promising individual sites, provides base datafor more detailed planning purposes and would be im-

11http://dbpedia.org/12http://sws.geonames.org/13http://www.ontotext.com/owlim/editions14http://wiki.gephi.org/index.php/

SemanticWebImport15https://gephi.org/


Raw dataset (format) Purpose Source

1 Global radiation (GRIB a) Calculate the algal biomass yield NCEP-NCAR 50-year reanalysisprovided by the CISL ResearchData Archive b.DSWRF (downward shortwaveradiation flux at the surface) indataset number ds090.2 [3]).

2 Temperature (NetCDF c) Calculate the algal biomass yield E-OBS dataset provided by theECA&D project d [10], [1].

3 Data on land use (GeoTIFF) Identify suitable land area European Environment Agency(EEA), CORINE land coverraster dataset 2006 version 13 e.CORINE land cover raster dataset2000 version 15 f.

4 Protected areas (GIS vector data) Identify suitable land area World Database on Protected Ar-eas (WDPA) provided on g byUNEP-WCMC [2].

5 Elevation (GIS raster data) Identify suitable land area Shuttle Radar Topography Mis-sion (SRTM) data [9] by GlobalLand Cover Facility (GLCF) h.

6 CO2 sources (MS Access) Identify sources of CO2 European Pollution Release andTransfer Register (E-PRTR) pro-vided by the EEA i in the versionof 09 Nov 2009.

ahttp://badc.nerc.ac.uk/help/formats/grib/bhttp://dss.ucar.educhttp://www.leokrut.com/leonetcdf/netcdfformat.htmldhttp://eca.knmi.nlehttp://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2006-rasterfhttp://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2000-rasterghttp://www.protectedplanet.nethhttp://www.landcover.orgihttp://bit.ly/v3dS2I

Table 1Sources of raw datasets

mensely useful to stakeholders in research, nationalcouncils and industry.

It is worth noting that the process of aggregatingdata for nutrients and water sources for each of thealgae production site is in progress. Once these be-come available they would be integrated with the otherdatasets as outlined further in Section 5.

4. Vocabularies

LEAPS utilises a set of several well established anddomain specific vocabularies as illustrated in Figure 3.

Spatial data has been modelled using a combinationof several ontologies namely, WGS84 ontology 16, spa-tial relations ontology, 17 the Geonames ontology 18

and the NeoGeo ontology 19.Geometries for algal plant sites and pipelines have

been modelled using an extension of the NeoGeo ge-

16http://www.w3.org/2003/01/geo/wgs84\_pos17http://www.ordnancesurvey.co.uk/

oswebsite/ontology/spatialrelations.owl18http://www.geonames.org/ontology/

ontology\_v2.2.1.rdf19http://geovocab.org/geometry


Fig. 3. Ontologies for algal biomass. Arrows indicate reuse

ometry ontology 20. For the CO2 sources, the geometryis modelled as a Point from the WGS84 ontology 21.

Modelling units and measurements for various at-tributes of the algal biomass datasets was non trivial.The QUDT ontology 22 for dimensions and units wasextended to include bespoke units of measurements.

We developed conceptual OWL ontology schemas23 for algal plant site, CO2 sources, regions andpipelines. Figure 4 illustrates some of the core con-cepts, their relationships and attributes. The figure alsoshows the relationship with the NUTS24 vocabulary.

5. The LEAPS datasets

The transformation process yielded four datasetswhich were stored in distributed triple store reposito-ries: Biomass production sites, CO2 sources, pipelinesfor the CO2 sources and region potential. We storedthe datasets in separate repositories to simulate the re-alistic scenario of these datasets being made availableby distinct and dedicated dataset providers in the fu-ture. While a linked data representation of the NUTSregions data 25, was already available there was noSPARQL endpoint or service to query the dataset forregion names. We retrieved the dataset dump. In order

20http://geovocab.org/geometry21http://www.w3.org/2003/01/geo/wgs84\_pos22http://qudt.org/1.1/vocab/dimensionalunit23Ontologies are available at

http:/purl.org/biomass/ontologies24

25http://nuts.geovocab.org/

Fig. 4. A partial account of core concepts, their attributes and rela-tionships

to inform the query retrieval performance, we prunedthe dataset to include only the regions in NWE. Wethen curated the pruned dataset in our local triple storeas a separate repository. The NUTS dataset was re-quired to link the biomass production sites and the CO2sources to regions where they would be located andto the dataset about the region potential of biomassyields. We further enhanced and augmented the NUTSdataset, with data on global radiation26. The trans-formed datasets, interlinked resources defining sites,CO2 sources, pipelines, regions and NUTS data us-ing link predicates defined in the ontology network de-picted in Figure 3. Figure 5 illustrates the linkages be-tween the datasets.

Fig. 5. Linked datasets for algal biomass

26global radiation is another term for solar radiation. It is the sumof short wavelength incoming radiation to the earth’s surface andconsists of the direct and the diffuse sunlight.


URIs

We propose URI patterns for the datasets used in thispaper to be reused across the sector. Note that while wewould like the URIs to be persistent, they may evolveas the uptake of linked data within the algal biomasscommunity gains momentum.

In particular we propose URI sets for

– Algal plant sites– CO2 sources– Pipelines.

Figure 6 exemplifies some of the URIs minted for realworld algal biomass entities which have unique identi-fiers and which are uniquely located in a certain NUTSregion. It also illustrates the definition of a conceptualentity and a relationship within the algal plant site on-tology.

Fig. 6. Representative URIs for Algal Biomass Plant Site

Statement level statistics

Statement level statistics for the various datasets inLEAPS are indicated in Table 2.

Dataset Number of Statements

Algae sites dataset 84703CO2 sources 14238Pipelines 261680NUTS 14469Global radiation 24738Biomass potential 4557

Table 2Dataset statement statistics

Dataset availability

The LEAPS27 application is available on the Web.The interface currently provides visualisation and nav-

27http://www.semanticwebservices.org/enalgae

igation of the algae cultivation datasets in a way mostintuitive for the phycologists. The application has beendemonstrated to several stakeholders of the communityat various algae-related workshops and congresses.They have found the navigation very useful and madesuggestions for future dataset aggregation. At the timeof this writing, data retrieval is relatively slow forsome queries because of their federated nature, how-ever optimisation work on the retrieval mechanism isin progress to enable faster retrieval of information.

Besides the application Web interface, datasets areavailable for querying via the dedicated triple storeWeb interface28.

VOID description of the datasets

The VoID29 descriptions of some of the datasets inthe LEAPS suite have been made available 30. Once thedatasets are made public the VoID descriptions will beupdated.

Linking to other datasets

Datasets within the LEAPS suite have been linked toeach other via the link predicates in the domain spe-cific vocabularies created for the datasets. The sitesdatasets have been linked to the CO2 sources, pipelinedatasets and the nuts region dataset. Back links to thedatasets have been provided from the CO2 and thepipelines dataset.

Though we have achieved good interlinking be-tween the datasets in the LEAPS suite, we have onlybeen able to link externally to DBpedia and Geonames.Reasons for the low external linkages have been dis-cussed in Section 7.

6. A Prototype Application

The LEAPS integrated datasets enables a screeningfor promising individual sites, provides base data formore detailed planning purposes and would be im-mensely useful to stakeholders in research, nationalcouncils and industry. We have developed a prototypeapplication 31 with a Web interface built over REST-

28http://www.semanticwebservices.org/openrdf-workbench/repositories/NONE/repositories

29http://www.w3.org/TR/void/30http://purl.org/biomass/void31http://purl.org/biomass/LEAPSDemo


ful Web services that exposes the LEAPS datasets viavarious facets. The Web interface provides an inter-active way to explore various facets of sites, sources,pipelines, regions, ontolgoies and SPARQL endpoints.Figure 7 illustrates a typical site. The map visualisa-tion has been rendered using Google maps. Besidesthe SPARQL endpoint and the interactive Web inter-face, a REST client has been implemented for access tothe datasets. Query results are available in RDF/XML,JSON, Turtle and XML formats. For the stakeholders

Fig. 7. A typical site as visualised with the LEAPS Web interface

in the biomass domain, the application provides an in-tegrated view over multiple heterogeneous datasets ofpotential algal sites and sources of their consumablesacross NUTS regions in NWE.

7. Discussion

While LEAPS currently provides integrated infor-mation about algal plant sites, CO2 sources and thepipelines connecting them, there are several otherdatasets such as nutrients, water supply and their as-sociated sources which need to be integrated oncethey become available. One of the core datasets whichshould be made available as linked data is that of algalstrains that can be cultivated on the plant sites. We haverecently curated the Algaebase 32 dataset as linkeddata. In the near future, experiments would be carriedout on the potential sites to establish the algal strainsthat can be cultivated there. The algal strains fromthe Algaebase dataset will then be integrated withinLEAPS to link the potential biomass production siteswith the algal strains they could produce. We believethis will go a long way in providing the stakeholders,

32http://www.algaebase.org

information about the kind of algae that can be culti-vated on potential sites, thereby helping in a more ac-curate analysis of the economic potential of producingbiofuels from Algae.

A limitation of LEAPS is the low number of out-going links it provides with other datasets. CurrentlyLEAPS links to DBpedia, Geonames and the NUTSdatasets. Three main reasons can be identified for theshortcoming in linkages:

– The lack of motivation within the algal biomasscommunity to open up and share data.

– The lack of shared vocabularies and uptake of Se-mantic Web and linked data technologies withinthe community. LEAPS is the first dataset suite tobe exposed as linked data using RDF.

– LEAPS is a newly curated dataset. Its availabilityas a data source to which other datasets can pro-vide outgoing links needs to be widely advertisedboth within and across the domain.

We strongly believe that the LEAPS dataset suitewill prove a major milestone in generating the muchneeded awareness about linked data and its benefitswithin the algal biomass community. We are workingwith biologists in the domain to address the above lim-itations.

8. Conclusions and Future Work

In this paper we presented a framework LEAPS thatexploits Semantic Web and linked data for making theanalysis of biomass potential in NWE available to thestakeholders. Specifically, the framework contributesby

– enabling the screening of data for promising indi-vidual plant sites and provides base data for moredetailed planning purposes.

– proposing a set of domain specific ontologies foralgal plant sites, CO2 and pipelines to be sharedand extended by the community.

– defining a linked data publishing architecture thattransforms raw data in disparate formats to a uni-form XML representation.

– using a set of well established and domain spe-cific ontologies as metadata to transform it furtherinto linked data.

– providing various data access options such as aSPARQL endpoint, an interactive Google map in-terface and a REST API for making the data ac-cessible to stakeholders.


As discussed in Section 7, several limitations need tobe overcome at various levels in order to fully realisethe vision of have an open platform for publishingand consuming algal biomass datasets, both within andacross community.

In order to increase the uptake and showcase the po-tential of LEAPS we have been presenting the applica-tion at various algae congresses and workshops. Thisalso informs us about any related datasets that can beintegrated within LEAPS. We are working with biol-ogists in the domain to facilitate the process of mak-ing the taxonomy from the AquaFuels33 project avail-able as SKOS models. Multifaceted visualisation ofthe integrated datasets is another area that we are cur-rently focusing on to motivate the idea of interlinkingdatasets. A few examples of these visualisations34 canbe seen via the Web application. The reasoning infras-tructure in LEAPS is currently based on implicit OWL235 DL inferences. Work is also in progress on exploit-ing rule based reasoning to model domain specific con-straints.

33http://www.aquafuels.eu/34Visualisations are powered by Sgvizler,

http://code.google.com/p/sgvizler/35http://www.w3.org/TR/owl2-direct-semantics/

References

[1] M.R. Haylock, N. Hofstra, A.M.G. Klein Tank, E.J. Klok, P.D.Jones, and M. New. A European daily high-resolution grid-

ded data set of surface temperature and precipitation for 1950–2006. J. Geophys. Res., 2008.

[2] IUCN and UNEP. The World Database on Protected Areas(WDPA). UNEP-WCMC, Cambridge, UK, 2010.

[3] E. Kalnay et al. The NCEP / NCAR 40-year reanalysis project.Bull. Amer. Meteor. Soc., 77:437–471, 1996.

[4] Teresa M. Mata, António A. Martins, and Nidia. S. Caetano.Microalgae for biodiesel production and other applications: Areview. Renewable and Sustainable Energy Reviews, 2010.

[5] Oilgae. Oilgae comprehensive report, energy from algae: Prod-ucts, market, processes and strategies. Technical report, Oil-gae, 2011.

[6] Claire Smith and Adrian Higson. Research Needs in Ecosys-tem Services to Support Algal Biofuels, Bioenergy and Com-modity Chemicals Production in the UK. Technical report,NNFCC, 2011.

[7] Monika Solanki, Johannes Skarka, and Craig Chapman.LEAPS: Realising the Potential of Algal Biomass Productionthrough Semantic Web and Linked data. In I-Semantics 2012:Proceedings of the 8th International Conference on SemanticSystems. ACM ICP Series, 2012.

[8] U.S. Department of Energy. National Algal Biofuels Technol-ogy Roadmap. Technical report, accessed June 2012.

[9] USGS. Shuttle Radar Topography Mission, 3 Arc Second sceneSRTM_ff03_nYYYeXXX (covering Europe), filled finished 2.0.Global Land Cover Facility, University of Maryland, CollegePark, Maryland, 2006.

[10] E.J.M. van den Besselaar, M.R. Haylock, G. van der Schrier,and A.M.G. Klein Tank. A European Daily High-resolutionObservational Gridded Data set of Sea Level Pressure. J. Geo-phys. Res., 2011.

linked data for potential algal biomass production monika solanki / linked data for potential algal...

Documents