Skip to content
Line Pouchard edited this page Mar 31, 2017 · 1 revision

Welcome to the ESIP Semantic Portal wiki! This wiki contains documentation for the ESIP Semantic Portal located here.

Motivation

Earth Science represents a diverse and complex data and metadata ecosystem that includes satellite observations and remote sensing, observations, experimental results, model data, and more. These data and accompanying metadata and descriptions are held in data centers that focus on types of data, such as ocean or ice data, hydrology data, climate model and simulation data. In order to perform analyses of specific phenomena, for instance predicting flash floods or visualizing a major perturbation like a hurricane, scientists need to integrate multiple datasets with different parameters at various resolution levels from different sources. In order to do this accurately, they need detailed descriptions and annotations of the chosen data sets, such as those provided by ontologies. Ontologies provide machine-processable entities organized in a graph that can describe these datasets so that datasets are more discoverable. Ontologies also provide labels and relationships between the entities so that ontology-annotated datasets can be discovered in multiple ways, by following relationships between entities, or by searching on related labels.

Problems we are seeking to address include improving the discovery and reuse of ontologies as they in turn support discovery and reuse of datasets. In Earth Science, most data and software are open source and available on the web. What makes these ontologies hard to find is the sheer volume of data, metadata, and annotations available from numerous sources. They often live on individual project web sites or in some cases in multi-purpose repositories such as Github. Increasing availability helps improve quality through feedback and reuse.

Ontologies are intended to be reused. Reuse improves the quality of annotations, since basic entities and relationships already exist and have been vetted elsewhere, for instance in the form of ontology design patterns. Yet, reusing entire ontologies can be problematic, because the purpose for which they were designed influences the semantic modeling of the covered domain. Finding mappings between ontology entities and linking ontologies through individual mappings allow ontology creators to reuse single concepts, rather than an entire ontology. Mappings enable data from independently constructed datasets to be related and combined.

Repurposing BioPortal technology

We have deployed an instance of the BioPortal Ontology Repository (the parent system) to host ontologies in Earth Science. This is reuse of a technology framework developed by the National Center for Biomedical Ontologies at Stanford University to host medical and biological ontologies. The framework contains many features including ontology validation, visualization, metadata, search and browse features, and a recommender. We are re-purposing the repository for Earth Science ontologies and adapting it for this domain. One advantage of our re-purposed repository is to provide a single site where Earth Science ontologies can be found, where researchers can share their ontologies, and build additional constructs on each other’s foundation. We have seeded the repository with some ontologies of interest in Earth Science. Our ontology collection includes the Semantic Web Earth and Environmental Taxonomy (SWEET) and the Environment Ontology (ENVO). ENVO contains 1911 classes. SWEET contains over 4,000 classes and covers many foundational sub-domains of interest to Earth Science, including basic physics and mathematical processes.

NCBO BioPortal Instance

NCBO BioPortal Technology is made available open source as a Virtual Appliance (VA). The ESIP Semantic Portal uses the latest available version of the NCBO Virtual Appliance: version 2.4 (April 2015) which contains the main features of NCBO Bioportal v. 4.15 (see release notes below).

The links and documents in this section are provided courtesy of the NCBO Bioportal documentation. NCBO maintains a wiki for working with the Portal and the VA at https://www.bioontology.org/wiki/index.php/Main_Page

BioPortal Virtual Appliance Release Notes are here: https://www.bioontology.org/wiki/index.php/BioPortal_Release_Notes

Organizational Structure

The ESIP Semantic Portal is currently deployed on the ECITE AWS instance as part of the collaboration between ESIP and ECITE. ECITE is an NSF-supported AWS cloud instance.