The Global Open Knowledge Hub: building a dream machine-readable world
The word ‘open’ has long been bandied about in development circles. We have benefited in recent years from advocacy to increase open access to research articles, and open data shared by researchers or organisations. But open systems that enable websites to talk to each other (e.g. open application programming interface) have been a little harder to advance into greater use, simply because they are not built for non-technical users.
The International Initiative for Impact Evaluation (3ie) recently joined eight other partners that are part of the new, DFID-funded and The Institute of Development Studies led Global Open Knowledge Hubproject to discuss several issues related to open systems. It was no surprise that all the partners spent quite a bit of time coming to their own understanding of an ‘open system’ and an ‘open hub’.
Put simply, the Global Open Knowledge Hub project will build an open system for sharing data between participating partners and with the wider world. As each of the participating partners offers knowledge services, there are thousands of research documents, articles and abstracts that are on our websites. To facilitate the sharing of these knowledge products, an open, web-based architecture will be built so that we can all just go to one place, i.e. the hub, and find high quality, diverse and relevant content on any chosen topic that is available from the partners.
To understand how the sharing works, step out of the human-readable world and step into the machine-readable world. If a machine can be programmed to search and read through the data, then the amount of data that can be processed starts to boggle the mind. The hub is a place where huge amounts of data in machine readable formats can be queried, accessed, used and combined with other data. If you are interested in climate change, one of the topics on which the hub project will focus, a huge amount of the research that exists on climate change, spread across continents, disciplines and sectors can be accessed in a matter of a few seconds. The sheer scale of it is awe inspiring. Think tons and tons of data, woven together in a kind of semantic web. This is what the web 3.0 world will look like.
All of this might sound like a grand vision. And as partners involved in pioneering work, we are aware that we need to get several things right for this vision to be realised:
As Edwards and Davies say in this paper, the current understanding of open data is primarily from the supply-side perspective. It’s not enough to just put out large quantities of data; we also need to get a better sense of the demand for the data. Who are our potential users? What kind of data would they need? What will they use it for? These are questions that need some serious investigation.
The IDS Knowledge Services Open Application Programming Interface is an example of a successful open system in the development sector. The Open Application Programming Interface (API) provides open access to tens of thousands of development research documents in its repository. According to Duncan Edwards, IT Innovations Manager at IDS, there is good demand for the IDS open API from both Northern and Southern development organisations.
But data has been accessed primarily via several applications – a mobile web application, regional document visualization application and a tag cloud generator. And these have been built to make the data accessible to non-technical users. So, we need more of this to happen to make the data in the hub more user-friendly and spur demand.
Get IT and content providers to work together
These open systems are not made for the ‘non-techie’, average user. When I first looked at the Open API of a website, the programming language that came up on the screen did not make any sense to me. But there is clearly a lot that the system can throw up for generating useful content for those with the technical skills to use it. For this to happen, researchers and communicators would have to work alongside a technical team and play a more active role in the curation of data. This is the only way the potential of the system can be fully explored.
Research is often labeled according to the needs and interests of its user. So the same piece of research may be tagged as agriculture development, rural development or farmers. In the machine-readable world, this becomes a crucial difference that prevents data on the same themes from linking to each other. The taxonomies we use to describe information change depending on the organisation, sector and country.
So for the hub, we need a system for classifying data, which maps these different languages to ensure that data on the same theme and topic can find each other and hang together.
Work out branding, attribution, licensing and copyright
How open can we be about sharing content? When our content gets used in some way e.g. featured on another website, will credit be given to the knowledge producer? If the knowledge service involves the production of summaries and abstracts of research articles, then it would be important to clarify with the original research producer on how they license others to re-use their content (e.g. creative commons).
Since research producers, knowledge service providers and funders often use web analytics as a metric for measuring success, organisations are often concerned that if their content is ‘open’, users may not ever visit their website. Thus they would be denied access to these important metrics. We need to therefore explore new ways of tracking how ‘open’ content is used beyond our own websites. Or we need to agree to share enough data so that users are directed to the originator’s website for full information.
The partners contributing to the Global Open Knowledge Hub are working through these issues. All the partners believe that development research has a crucial contribution to make to poverty reduction, but only if it is easily available and quickly accessible to users. So, what we are building together needs to become the prototype of what open systems should look like.
This blog post first appeared on The Institute of Development Studies’s Impact and Learning blog site.
Add new comment