Judges’ Queries and Presenter’s Replies

  • May 20, 2013 | 07:25 p.m.

    who will populate this ontology? Will this be automated from some data or manual?

  • Icon for: Melinda Neville

    Melinda Neville

    Presenter
    May 20, 2013 | 10:28 p.m.

    Thanks for the excellent question. There are Protégé plug-ins like DataMaster that will import from a relational database. In the first phase of my research, data from many Hg monitoring projects were collated into a geospatial database. Some of the data relevant to Maine are shown on a slide during the video (timestamp 2:02). Spatial and temporal relationships were evaluated (timestamp 2:26) using standard and spatial statistics, but the heterogeneity of the data limited our ability to form a cohesive picture. Thus, I proposed an ontological approach to the data. The ontology will be constructed and tested using my current database, followed by a series of research community workshops to further test and refine the ontology.

  • May 21, 2013 | 08:35 p.m.

    How much confidence do you have that the researchers and stakeholders will indeed use the common terms and controlled vocabularies developed in your ontology? There must be a large number of technical reports and publications in scientific journals and conference proceedings on all aspects of mercury monitoring, and it is normally hard to change the technical terms used by certain communities and replace them with a new ontological system. Could you comment on this issue?

  • Icon for: Melinda Neville

    Melinda Neville

    Presenter
    May 22, 2013 | 10:07 a.m.

    You’ve highlighted one of the greatest challenges- not only for ontologies, but also for broader scientific data management. Because incomparability of Hg data across projects is a well known limitation to spatiotemporal trend determination, there has been broad acknowledgement of the need for more robust tools for data analysis, archiving and sharing. However, the goals of our research at this stage are not to replace current data collection methods, but to facilitate collaboration between diverse Hg research communities.

    The process of collating the relational database for this project was primarily an exercise in metadata creation and preservation (units, analytical methods, geolocation, etc). The Hg Research Ontology will provide a platform to query explicit and implicit relationships between these heterogeneous data, and its efficacy at handling diverse metadata will have to be proven. An experimental portion of this research will be the testing the competence of the ontology (please refer to the Development section of the poster), which must include the ability to reason with disparate data. In order to be useful and have the opportunity to expand, the Hg Research Ontology must be robust enough to incorporate old and new data. If successful, there will be an incentive for researchers to use the ontology and explore how their data fits into the bigger puzzle.

  • May 21, 2013 | 09:39 p.m.

    How are academic researchers, governmental agencies, and other stakeholders being involved in the process of developing the ontology?

  • Icon for: Melinda Neville

    Melinda Neville

    Presenter
    May 22, 2013 | 10:19 a.m.

    Maine is a small state with a big Hg problem- the agencies with research related to Hg are well integrated. One of this project’s primary investigators is from the Maine Department of Environmental Protection, and much of the legacy data here is from Acadia National Park, thus the National Park Service is a partner as well. Data from other sources such as the Biological Research Institute (www.briloon.org) were acquired with the agreement that any results from the project would be shared. In our latest grant proposal, we received funding for a workshop to bring these researchers together to further test, dissect, and challenge the ontological framework. Prior to that workshop, I will have the ontology available online for exploration and comment. Without this broad support, the Hg Research Ontology could not become a useful tool.

  • Further posting is closed as the competition has ended.

Presentation Discussion

  • May 22, 2013 | 11:22 p.m.

    This is a great research question! Do you know of other similar examples of ontology research for environmental problems?

  • Icon for: Stacy Doore

    Stacy Doore

    Trainee
    May 23, 2013 | 08:52 a.m.

    Michael, I am also a PhD IGERT at UMaine in Spatial Information Science and Engineering. Besides my own masters work on an ontology to capture cumulative exposure histories in human populations representing longitudinal mobility and multidimensional factors, there are a number of emerging ontologies to capture data like this for specific environmental domains. Are you interested in a specific environmental area or species domain?

  • Icon for: Melinda Neville

    Melinda Neville

    Presenter
    May 23, 2013 | 11:28 a.m.

    Thanks, Michael!
    Yes, there are several examples of related research. The Marine Metadata Interoperability Project (marinemetadata.org)has built a semantic framework for managing marine science data, which are as varied and more numerous than Hg data. Ecological data management strategies are the focus of the National Center for Ecological Analysis and Synthesis (www.nceas.ucsb.edu), including the development of the Extensible Observation Ontology for synthesizing ecological data. These examples provide a jumping off point for developing a multifaceted approach to Hg data management. The first ontology that I am currently working on builds the computer-readable biogechemical cycle for Hg in freshwater ecosystems. A second, integrated ontology will incorporate the observations of Hg and their links to specific aspects of the Hg biogechemical cycle.

  • May 23, 2013 | 04:03 p.m.

    Thank you both for the helpful replies. Stacy, I’m not interested in anything in particular at this point but will keep in mind the projects and resources you both bring to light here. As Melinda says, this looks like a good jumping off point for me to look further into it.

  • Icon for: Stacy Doore

    Stacy Doore

    Trainee
    May 23, 2013 | 08:48 a.m.

    Melinda, great job on the video of your work! Are you planning on formalizing your ontology in descriptive logic or will it be just a conceptual framework?

  • Icon for: Melinda Neville

    Melinda Neville

    Presenter
    May 23, 2013 | 11:31 a.m.

    Thanks, Stacy-
    I plan on formalizing the ontology- but am still in the planning and discovery phases. What do you suspect the biggest challenge of making that leap will be?

  • May 23, 2013 | 04:16 p.m.

    Hi Melinda,

    Nice job on the video! Sorry I have been tied up in meetings the last few days so did not have a chance to visit earlier.
    Kate

  • Icon for: Erik Koepf

    Erik Koepf

    Trainee
    May 24, 2013 | 09:53 a.m.

    Great topic and well explained, nice work!

  • Small default profile

    Dave )

    Guest
    May 24, 2013 | 02:57 p.m.

    Great topic, and in many ways a nicely constrained problem that should be conducive to the use of ontologies. Rather than creating the “ontology to end all ontologies”, you have a focused – and very important – area, and it seems a great deal of data to experiment with. This seems like it has the potential to make a real contribution to an important area of research in environmental science, as well as in the application of ontologies to a significant scientific problem.

  • Further posting is closed as the competition has ended.

Icon for: Melinda Neville

MELINDA NEVILLE

University of Maine
Years in Grad School: 5

Mercury (Hg) Research Ontology: Employing Informatics in Geochemistry

Mercury (Hg) monitoring in the United States has spanned three decades, but the diverse nature of Hg research makes comparison across studies difficult. The ability to measure whole ecosystem changes in Hg contamination in response to policies regulating Hg is often limited, and researchers have difficulties comparing results or utilizing the wealth of legacy Hg data more broadly. Commonly, there are several methodological and contextual barriers to comparing disparate monitoring efforts. For example, the differences in study design between sampling and monitoring projects may be in scale, duration, or intensity of sampling. Likewise the sampling methods, laboratory procedures, and reporting further complicate the evaluation of multiple monitoring efforts. For Hg contamination, there is often an assumed common understanding of some terms, assumptions about relationships, implicit, or different specification of the landscape settings, and imprecise or ambiguous spatial context specification of observation units that hinder the ability to make logical linkages between study results. These relationships and supporting data connections to these relationships can benefit from more formal definition. A Hg Research Ontology will provide a way to explicitly capture knowledge about the specific domain, and support consistent and unambiguous representations of entities and relationships within the field. In this research, we are developing ontologies to facilitate disparate data integration, dissemination and comparison for Hg monitoring in freshwater ecosystems. The developed ontologies will allow Hg data to be placed in the context of the Hg biogeochemical cycle and linked to contextual characteristics of the observation settings.