Integration and Data Management for the Feedbase Investment Plan

Summary

Research projects are increasingly expensive and ways of getting more value for meat producers from the current suite of feedbase projects are needed.  One way to extend the value of FIP research projects is to make sure that they either use or contribute to the existing grazing systems models (GrassGro, SGS and MIDAS) to allow the exploration of 'what if' questions for locations and scenarios that were not included in the experimental program.  Another mechanism to increase the value from research sites is to store the data centrally and make it available in the future for different researchers or modellers, doing different analyses or asking different questions. 
The specific objectives of this project were - By 15th June 2013:
1.   Developed guidelines for:
a. Minimum data collection at each FIP research site
b. Protocols for standardised data collection, storage and meta-data
2.   Make recommendations for a centralised FIP database
3.   Make recommendations for minimum data to be collected at FIP Participatory R&D sites
4.  Based on Objective 1 above, identify key gaps in the existing data collection protocols at existing FIP project sites
Each of these objectives have been met, and they are individually discussed in Sections 5.1 to 5.5.  Most of the challenges associated with this project related to objective 2 – make recommendations for a centralised FIP database – and this objective dominates the discussion in this final report.
Extensive consultation with MLA, modellers, FIP project teams and IT specialists formed the basis for the recommended process presented in this report.  It follows an interim report where 3 possible technical options were presented to MLA in the Milestone 2 report (Appendix A).  MLA responded with the request that the final approach be designed to deliver as much of Recommended Option 2 as possible but with as much of the simplicity of the simpler Recommended Option 1 as possible.  This has been achieved and was tested with research leaders from the FIP projects via a discussion paper and at a workshop in Canberra.
In summary, the data management system is not a relational database but it does provide MLA with a system to collect, store, find and retrieve data from all FIP projects for unknown future uses by unknown future users.  The system will use SharePoint, (from Microsoft) that is being rolled out across MLA for other purposes and therefore it is not an additional requirement for the FIP.
The recommended system provides capability for:MLA to collect and store all FIP project data from field experiments, with some minor exceptions (eg the massive datasets associated with gene sequencing projects)Significant search capability (ie findability) across all data assets and project reportsSearching is based on metadata values and automatically generated classification categoriesMigration to a more sophisticated relational database storage approach at a later date if that should be required.
The recommended system does not provide capability for:Automated cross site analysis, for which a relational database would be required but the FIP projects have not been designed to allow such cross site analysis.  Cross site analysis would require data retrieval and followed by the manual construction of the analyses.'Findability' does not extend to retrieving cross-project-information based on filtered metadata ranges such as selecting projects that had a soil pH between 5.5 and 6.5 or that fall within specified geographic boundaries.  This must be done manually.
The approach is low in solution complexity and is the least disruptive to existing organisational practices for MLA, but with the requirement for participating research organisation stakeholders to supply a minimum data set, raw project data, and the metadata needed to ensure the raw data can be correctly interpreted and used.
Additional compliance requirements for MLA personnel over and above existing workflows, protocols and procedures already in place to manage research projects are small.
It steers the middle road between simplicity (to maximise compliance) and functionality (to maximise 'findability') while staying clear of many of the risks and complexities associated with this type of system development.
Included in the report is a focus on easy implementation to minimise non-compliance, and a draft set of rules around which MLA and the FIP Steering Committee can begin a discussion to govern the future use of project datasets.  IP issues remain a challenge that MLA will need to negotiate with FIP partners should the recommended processes in this report be adopted.