Page History
| Warning | ||
|---|---|---|
| ||
The eagle-i software is still available to the open-source community, but is no longer supported or maintained. |
| Table of Contents | ||
|---|---|---|
|
Introduction
eagle-i is a distributed platform system for creating and sharing semantically rich data. It is built around semantic web technologies and follows linked open data principles. In its current incarnation and operational deployment, eagle-i focuses on biomedical research resources. However, thanks to its ontology-centric architecture, the platform can be adapted to other domains.
The An eagle-i platform comprises software components deployed at a participating institution (an _eagle-i node) _and network comprises eagle-i nodes installed at participating institution, and a central search application. Figure 1 provides a high-level overview of the eagle-i software components of a node:
Figure 1. Overview of the eagle-i software stack
eagle-i nodes are independent of each other. The central search application provides a unified view of the data across nodes. eagle-i resources are captured and stored as as RDF data data that conforms to a resource resource ontology. At the core of the eagle-i node software stack is an RDF repository that provides services to institutional data collection applications, to an institutional search application and to the central search application. Data collection applications exist for both manual annotation and bulk data processing.
The The eagle-i ontology drives drives the data collection and search user interfaces and internal mechanisms, and is used for structuring and validating data in the repository and for indexing resources. This design choice allows applications to seamlessly adapt to ontology evolution, and provides ontology developers with a mechanism to rapidly test and refine their models. Figure 1 depicts an eagle-i network at a high level.
Figure 1. Architecture of an eagle-i network
The eagle-i software stack
Figure 2 shows a more detailed view of the an eagle-i node's software stack components.
Figure 2. Details of the eagle-i stack
Repository
Wiki Markup
Data tools
The eagle-i platform utilizes a variety of tools for repository data ingest, broadly referred to as data tools. The different tools share a common layer for interacting with the eagle-i repository.
The SWEET (for Semantic Web Entry and Editing Tool) is a web application for manual data entry and curation developed using the GWT (Google Web Toolkit). Its core component is a dynamic forms generation module that translates ontology axioms into UI data entry widgets. The SWEET generates a form per ontology class and thusly allows users to create instances of an ontology class. Navigational elements of the SWEET include workflow controls and instance listing and filtering.
An The SWIFT ETL (Extract Transform and Load) toolkit provides command line tools for transforming Excel and XML tabular data into eagle-i linked instances, and for loading them into an eagle-i repository. The data mappings necessary for performing the transformations are captured in an RDF map, in an approach heavily influenced by the RDF123 tool.
...
The search application backend is a semantic search framework, where pluggable search providers perform searches against different types of indices and public services. A Solr search provider, the core backend component of the search application, builds a collection of Lucene/Solr indices that enable search and autocomplete functionality across the eagle-i dataset and ontology. The search backend also includes providers for a variety of external sources such as NCBI and NIF.
The search application front-end is a web application developed using the GWT (Google Web Toolkit). The search UI includes features such as faceted search, synonym expansion, autocomplete based on instance data and ontology terms.
...
The search application and the different data tools use a data model component, the EIOntModel, to obtain information about ontology classes and properties. The data model is optimized to be used by client-side applications.
In March 2012, the eagle-i ontology was officially released. For more in-depth information about it, see the google code project site.
The eagle-i network
In the eagle-i network, institutional servers are independent of each other. A central search application provides a unified view across the institutional servers.
We have experimented with a deployment where the central search application queries institutional search indices using SPIN, a federated query network that distributes queries and aggregates results. This provides maximal institutional independence, as institutions can join and pull out of the federation at any moment. However, with an heterogeneous data set such as eagle-i's, ranking search results is a non-trivial task.
We are currently using an alternate deployment whereby the search application builds central indices by harvesting data from the institutional repositories.
Figure 3 shows an eagle-i deployment using SPIN.
Figure 3. Sample eagle-i deployment
Terminology
| Wiki Markup |
|---|
*\[1\] eagle-i resource instance *\- a collection of RDF statements about the same subject or about an _embedded instance_ \[2\] subject, plus the display labels of all predicates and objects used in these statements. |
Figure 3 illustrates our ontology-driven approach.
Figure 3. Ontology-driven approach
Terminology
[1] eagle-i resource instance - a collection of RDF statements about the same subject or about an embedded instance [2] subject, plus the display labels of all predicates and objects used in these statements.
[2] embedded instance - an instance that can only exist in the context of a parent instance and that cannot be linked to from other instances. Embedded instances do not have provenance Wiki Markup
How are we doing?
Is there anything that could be clearer in our documentation? We welcome your questions and feedback.
...
Overview
Content Tools