6.4 Data standards

The lead grantee must decide how data standards are relevant to this project. This will depend on the type of project.

• Is the project building on previous research, and is it wanting to add an additional perspective of knowledge into an existing body of work? For instance, say that a crop is to be grown and tested under a new set of conditions to complement previous research. In this case, the Programme Manager (PM) might want the data standards to be rigorously defined and structured. A good tool in this case is the Agronomy Ontology (AgrO), where the logic and data elements are defined, and into which new research can find place in relation to past findings. In this case, it is good to start from a predetermined ontology, so that the output of the project adheres to clear data standards that allow for integration of results.

• Is the project taking a more creative step open to new understandings coming from the research? Is the project seeking to examine a new question that needs fluidity to come up with answers? An example of this is Waterways, where the complexity of inputs into farmer crop yields needs a blank canvas to be investigated. This requires an openness not to fit the data and logic into an existing structure but to let the data identify the pivotal concepts in the problem. In this case, it is important that the researcher chooses a more open set of data standards, such as arranging data on a topological map, in order to draw conclusions. The data standards are still important in making sure the input from the datasets are consistent, but should not predefine the scope of the research.

In each of these cases, data standards play a critical role, requiring different handling to best fit the analysis into accessible and shareable outcomes.

Data standards help form the course of research and set the foundation on which the processing of data can be securely managed.

• In the case of a very familiar context for the research, data can be fitted into an existing logic and model that immediately joins the processing and the results into a larger body of knowledge. A good tool in such cases is Global Agricultural Research Data Innovation Acceleration Network (GARDIAN), developed by CGIAR, which will guide all the steps of data manipulation in the vocabulary of AgrO as foundation. At each stage of the project, the tools, data and processing steps can be checked to make sure the experiment can interchange its results with peers. In this way, the research is at once relevant to a wider debate, and can build up knowledge across a wide range of research projects.
• In the case of research seeking to make totally new understandings about a region or inquiry, the data standards need to be much looser. One does not know in advance the logical structure into which the data will fit. The data standards are now more focused on maintaining integrity across the different data providers contributing to the project, to realize an internal consistency of data use. This is the case in Waterways, where it is only at the end of the research that the pivotal relation of water availability to crop health and ground temperature makes itself known. The data foundation is in this case the map of GIS, upon which the data comes together into a whole picture distinctive to the area.

Data standards are vital in integrating different users’ ways of storing information. A well structured data standard will also be extendable if new fields have to be added, or if compared to outside datasets. However the data situation might change in the future, the data standard gives a reference that can hold the whole picture of what is happening. This creates a foundational structure into which data can be embedded, and presumes a logical form for analysis.

• AgrO takes much of the headache of handling data out of the hands of the researcher. Once you make the mapping from your current research to the existing template of definitions, the associated GARDIAN toolkit can navigate the data value chain reliably and routinely. Much of the difficulty of engaging with data is incorporated into standard tools that are ready to be applied at each stage of the process. The data adheres as far as possible to standards from other work, so that the results are at once applicable and shareable with related conclusions in similar trials. The ontology naturally steers the research to comply with the data standards chosen.
• If researchers want a more flexible approach to how the data interprets the situation, they need to pay more attention to complying with data standards, as mistakes can easily happen. For instance, in Waterways, the village names are listed in advance to avoid ambiguity when asking the farmer where they are from. KML (Keyhole Markup Language) is chosen as a standard format for all satellite and field data as this format is widely recognized and can be easily imported into geographical map programs. The geographical data model used has its own pre-set data standards that the project must honor.