TERN Eco-informatics met a key milestone last month when it quietly released the first beta (pre-production) version of the data portalof the Australian Ecological Knowledge and Observation System (ÆKOS).
The release depends on completing the complex information framework and architecture model of ÆKOS – the ‘back end’ – that underpins the portal, and is an entree into the ‘look and feel’ of the forthcoming production version. As it stands, the beta version is under active development, with Eco-informatics engineers using an approach that suits the R&D nature of much of the leading (some would say ‘bleeding’) edge activities required to build such an innovative e-research application.
AEKOS R&D approach
The spiral model of application development defined by Barry Boehm is probably the closest approach to the engineering perspective that the Eco-informatics team has adopted to develop ÆKOS. It is particularly suited to complex and novel engineering tasks where the requirements are uncertain and ill-defined: specific knowledge and understanding evolves and matures during the design and implementation phases, and engineers use this to build a good solution for each component of the system. ÆKOS is a suitable candidate for this approach.
The Eco-informatics team has adopted the principles of iterative and incremental engineering underlying the spiral model to manage the constraints and implementation risks.
Their approach also ensures that complex components of the architectural framework and important portal features can be reviewed by data users, with feedback embedded in later iterations until the most complete version is produced.
Therefore, Eco-informatics focuses on an adaptive approach. A number of steps are applied, in the following order, to engineer the ÆKOS components such as information models, data ingestion scripts, the data repository and the search engine.
- Step 1. Identify the research infrastructure problem for ecological data.
- Step 2. Review existing and related infrastructure solutions to scope the complexity and levels of innovation involved.
- Step 3. Design the information framework and architecture model.
- Step 4. Incorporate feedback on the design from the e-research and ecosystem science communities.
- Step 5. Build a proof-of-concept portal (alpha release) and subsequent iterations (beta releases), leading to a production release, and perhaps further production releases with additional capability.
- Step 6. Seek feedback on each release from the e-research and ecosystem science communities.
- Step 7. Incorporate feedback and learnings into subsequent iterations, repeating steps 3 to 6 as appropriate.
With the foundations complete, the focus has moved to the portal.
Aspects of the spiral model adopted to develop the ÆKOS data portal. Four phases are described, with iterations through each phase to produce progressively refined versions of the portal (alpha, beta ‘1 to n’ version) until the first production version which is due in May. (Image modified after Bohm)
Purpose of the portal
The data portal represents the interface the user has to the underlying information framework and architecture model. So building the portal involves implementing the components that allow users to interact with the data. The components are ‘stitched together’ to represent the full functionality or feature-set of the data portal, and provide an efficient means of understanding the full richness and complexity of ecological plot data.
The portal has several key features: search, map interface, visualisation and extraction.
The search functions let users browse and question the ÆKOS data repository by indexed search terms to accurately retrieve a set of results. This may be done via a query builder enabling complex, nested search expressions to be authored or by using a map interface.
The map interface displays a range of contextual data layers while allowing users to identify, select and view study locations and highlight search results.
Visualisation provides ways of exploring the complex data in ÆKOS including:
- an ‘Observation Graph Viewer’ to show the relationships between different data elements at a location;
- a ‘Process Graph Viewer’ to show the relationships between the methods and other activities at a study location, and also to display a full description of all methods used to collect the data;
- a suite of linked documents that describe key components of individual surveys, similar to metadata but with more context and structure;
- a ‘summary description’ template for each search result, which contains details of ownership, access, licensing conditions, and spatial, thematic and temporal coverage.
Data extraction allows the user to select a set of results provided via the search capabilities of ÆKOS and add this to a shopping cart to download and analyse off-line. Before they can download records, they must agree with the licensing conditions for each dataset.
The Director of the Eco-informatics facility, Mr Craig Walker, said the team still had some way to go before the portal would sufficiently reflect their thoughts on what should be available in the production release.
Calling on ecologists for feedback
‘In the interim we really need the ecological community to use and review the portal and start a dialogue with our Data Facilitator to report issues and, more importantly, to work through “beta useability” to make sure that the team addresses the research-data needs of ecologists,’ Craig says.
‘Although we have moved rapidly along the development path towards laying the foundations for the “production” portal, we still need to activate all of its features (or functionality) and smarten up its appearance. “Component stubs” have been set up as place-holders for all the additional features that we intend to implement.
‘For example, we’ve built a shopping basket to enable data extracts and support licensing, but we haven’t yet implemented the mechanism to fill the shopping basket contents, hence the notion of a beta versus a production release. We’ll implement refinements quite regularly and publish them in the “What’s new” pop-up window over the next few months. I suggest users check for new updates before getting started on the portal.
‘The beta portal has been released publicly because it is imperative that the e-research and ecological communities use and evaluate it so we can improve the user experience and ensure that this becomes a valuable tool to support ecologists.’
The Eco-informatics team will present and run demonstrations of the portal at the TERN Science Symposium. They will be seeking feedback on the portal and will be available to discuss their work.
For more information, contact Craig Walker or the Eco-informatics Data Facilitator, Dr Anita Smyth.
Published in TERN e-Newsletter January 2013
