TL;DR

Come and talk about developments in the application of machine learning in open data and international development, then maybe join us for a drink afterwards!

IATI miniTAG

The International Aid Transparency Initiative (IATI) is an open data standard concerned with international development and humanitarian expenditure combining data published by nearly 600 national governments, multilaterals, civil society and private organisations worldwide. A 'miniTAG' is the effectionate name given to an informal meeting organised by members Technical Advistory Group (TAG). It's not all about IATI, but there will be stickers!

Theme

Machine learning (ML) permeates modern life. The advances it has brought to advertising and customer profiling are well documented, but examples of it bringing similar benefits to open data and transparency are rare.

There is a much more sober side to machine learning, less fantastical than AIs winning games of Go, or Google’s servers dreaming up other wordly images.

Namely there are sets of complex problems which are very hard to solve logically, but which can have optimal answers found - these are optimisation problems, and there are many within the world of open data and international development.

Open Ag has been taking two of them on recently. Given only the description text and some unstructured supporting documents for an agricultural investment in IATI:

How can you automatically assign a detailed agricultural code?
How can you automatically assign a detailed location code?

We are in the process of developing and testing some ML tools to try to find optimal solutions to these problems (and there will be a short demonstration of these tools in the event, and an opportunity to play with them for those who are interested).

The aim of this event is for members of the IATI community and beyond to discuss the potential benefits and challenges presented in these kinds of problems and solutions, drawing on their own experience of open data publication and use.

Beyond presentations, we are hoping to consider the following questions:

What other optimisation problems exist in IATI and open data more broadly?
What are the limits to the use of machine learning within the publication of open data, and within its use?
What kind of common principles should we ascribe to when developing this kind of technology, and when using it?

Structure

Following lessons learned from the first miniTAG, this event will begin with a short introductory talk, followed by a semi-structured discussion of the questions raised, informal breakout discussions and networking.

We are expecting a mix of representatives from bilateral donors, multilateral organisations, civil society and the private sector, and we want people to have a good opportunity to socialise; not every discussion has to end with a deliverable, or future action!

The introductory talk will be given by representatives from the Open Ag Funding partnership and will introduce attendees to the concepts mentioned above, as well as the applied example of the work being done by the Open Ag Funding Partnership.

IATI miniTAG Session: Machine Learning in Open Data & Development