Integrating Ecoinformatics Resources on the Semantic Web

Cynthia Sims Parr

Andriy Parafiynyk

Joel Sachs

Li Ding

Sandor Dornbush

Tim Finin

Department of Computer Science and Electrical Engineering

University of Maryland Baltimore County

Baltimore, MD 21250 USA

David Wang

Allan Hollander

Dept. of Computer Science

University of Maryland

College Park, MD, 20742 USA

Information Center for the Environment

University of California Davis

Davis, CA, 95616xxxx USA



We describe ELVIS(the Ecosystem Location Visualization and Information System), a suite of tools for constructing food webs for a given location. We express both ELVIS input and output data in OWL, thereby enabling its integration with other semantic web resources. In particular, we describe using a Triple Shop application to answer SPARQL queries from a collection of semantic web documents. This is an end-to-end case study offor the semantic webs utility for ecological and environmental research.

Categories and Subject Descriptors

H. Information Systems

H.4 Information System Applications

H.4.m miscellaneous

General Terms

Algorithms, Design, Experimentation, Human Factors, Standardization,


Food webs - Ecological forecasting - Semantic web - Ontologies - Invasive Species - Biodiversity - Service Oriented Design


SPIRE (Semantic Prototypes in Research Ecoinformatics)- is a distributed, interdisciplinary research project tasked with building semantic web prototypes for invasive species science. {SPIRE: Semantic Prototypes in Researchh Ecoinformatics 2006 #6010} .

Our main integrating suite of tools is ELVIS (the Ecosystem Location Visualization Information System). ELVIS is motivated by the belief that food web structure plays a role in the success or failure of potential species invasions. Because very few ecosystems have been the subject of empirical food web studies, response teams are typically unable to get quick answers to questions like what are likely prey and predator species of the invader in the new environment?

The core data has been integrated from publicly available sources and is now available on the semantic web. We have constructed a platform for investigating multiple algorithms for food web prediction. Further, by exposing item-level data through several rich sets of ecological and evolutionary ontologies, and by providing these tools as web services, we enable integration with other semantic web/web 2.0 applications, such as SwoogleWOOGLE. WE and FieldMarking {Parr #6020} . We have developed a shopping cart application, Triple Shop, which allows a user to select semantic web documents, and to issue SPARQL queries over their union. Thus, we are able to integrate diverse ecoinformatics data in response to ad-hoc queries.

1.1 Related Work

Previous work on data integration in ecological informatics includes online data repositories[2] and workflow [4] ontologies. Metadata allows only the discovery of possibly interesting datasets and does not provide the means to harvest the data itself. Individual food web researchers maintain and share their own digital data archives, in individualized data formats, though more accessible standardized archives are beginning to emerge[1]. There are goodexcellent databases on invasive species {Invasive Species Specialist Group 2006 #6060} ]s (e.g. but they areit is not automatically integrated with information about non-invasive specieswith which they interacywith which they may interact;, nor is there web-based support for modeling an invasive species, anywhere.


The task of providing food web information for a user-specified location breaks into two distinct problems: constructing a species list for a given location; and constructing a food web from a given species list (and habitat information).

Figure 1. Nile Tilapia, an invader in Florida ecosystems, is predicted to eat algae and have no potential predators. Organisms predicted to be nearby in the food web (to the right of algae) could be impacted by or mediate the introduction of this competitor.

2.1 Species List Constructor

Our goal is to allow a user to input a location, and get back a species list for that location. This is a hard problem, typically ad-hoc, and relying on expert knowledge. There are, in general, three kinds of information that can be used to generate a species list: (i) park inventories; (ii) point locations, e.g. from specimen descriptions in museums and herbariums; and (iii) distribution maps generated by applying statistical techniques to point locations. We are integrating all of the above for California, and expect that the ontologies and synthesis strategies we have developed will apply to other states, and enable ELVIS to quickly spread beyond California.

2.2 Food Web Constructor

The Food Web Constructor (FWC) uses empirically known food web links to predict food web links not yet recorded.

Copyright is held by IW3C2.

WWW 2006, May 22-26, 2006, Edinburgh, UK.

A user can choose which food web studies to use for prediction or exclude from 257 datasets we compiled from previously digitized literature. {Cohen 1989 #6030} , {Vazquez 2005 #6040} ,{Dunne, Williams, et al. #6050} .


Taaxa can be entered several different ways: simple text lists, XML files, or food web number. In this latter case we seek to reconstruct feeding links based on the rest of the database and can therefore assess the success rate of the different algorithms or model parameters.

Each suspected link is reported, together with references to supporting evidence. Summary statistics of the food web are also reported.

2.3 Evidence Provider

Figure 2 shows the evidence for a predicted trophic link - namely, the actual link that was observed, the study in which it was published, and the relationship between the species in the observed link and the predicted link.

2.4 Technical Approach

The Species List Constructor interacts with web services that provide a variety of species informatione top half of ELVIS uses ontologies (how?). OurThe current Food Web Constructor algorithm uses taxonomic distances to weight evidence supporting or failing to support links between organisms. All data input, output, and taxonomies for Food Web Constructor and Evidence Provider are available in OWL on demand. , but at present the Ccalculations are performed act on data residing in MySQL databases. Scripts generate OWL documents from Animal Diversity Web {Myers 2003 #5080} ( via the ETHAN ontology. The triple shop is currently implemented using the joseki( .Add Joseki web service and

3. Swoogle and Triple Shop

Swoogle (Google for the Semantic Web) is ourthe semantic web search engine [3](Swoogle ref). It allows users to search for both ontologies and instance data (collectively referred to as semantic web documents) along a variety of parameters. Once documents are returned, a user can select certain of them for inclusion into her Triple Shop, a sort of shopping cart for RDF triples. We have built a stand-alone version of the Triple Shop, which allows a user to specify the URLs of arbitrary semantic web documents, and to issue SPARQL queries against the union of those documents.

3.1 Using the Triple Shop to Integrate Food Web and Natural History Data

ELVIS illustrates the potential of the semantic web to support rapid querying of distributed scientific databases for a variety of scenarios. For example (Figure 3): Determine known predator-prey relationships among an invader and a specific group of native species in a particular habitat, as reported in previous studies.

Figure 3. SPARQL query on ETHAN and SpireEcoConcepts OWL documents.


This research was supported by NSF ITRgrant #0326460 and matching funds received from the USGS Nat.ional Bio.logical Information Infrastructure.


[1] Dunne, J. A. The network structure of food webs. In: Ecological Networks: Linking Structure to Dynamics in Food Webs, eds. Pascual, M. and Dunne, J. A. Oxford University Press, 2005.pp. 27-86.

[2] Jones , Matthew .B.; Berkley, C.had; Bojilova, J.Jivka; Schildhauer, M. ark P. 2001. Managing scientific metadata. IEEE Internet Computing. Vol: 5(5). Pages 59-68.

[1] Li Ding et al., "Swoogle: A Search and Metadata Engine for the Semantic Web", Proceedings of the Thirteenth ACM Conference on Information and Knowledge Management, November 2004.


[4]Ludaescher, B. et al. 2004. Scientific workflow management and the Kepler system. Concurrency and Computation: Practice and Experience. Ludaescher, B. Bertram; Altintas, Ilkay; Berkley, Chad; Higgins, Dan; Jaeger, Efrat; Jones, Matthew B.; Lee, Edward A.; Tao, Jing; Zhao, Yanget al.In press 2004. Scientific workflow management and the Kepler system. Concurrency and Computation: Practice and Experience.Anderson, R.E. Social impacts of computing: Codes of professional ethics. Social Science Computing Review, 2 (Winter 1992), 453-469.


Conger., S., and Loch, K.D. (eds.). Ethics and computer use. Commun. ACM 38, 12 (entire issue).

Mackay, W.E. Ethics, lies and videotape... in Proceedings of CHI '95 (Denver CO, May 1995), ACM Press, 138-145.

Schwartz, M., and Task Force on Bias-Free Language. Guidelines for Bias-Free Writing. Indiana University Press, Bloomington, IN, 1995.