The BilotGo hackathon concluded last June with the AI-boosted auditing tool, Risk Detector, taking first place. In brief, Risk Detector utilizes network science and machine learning to highlight various perspectives relating to government procurements and their inherent risks, thus enabling auditors to shift their work from tedious manual inspection of data to being able to explore curated points of interest relating to their work.
Risk Detector goes to Peru
As Risk Detector enters a piloting phase at the National Audit Office of Finland (NAOF), the tool has also enjoyed worldwide attention among similar national audit entities in countries all over Europe and elsewhere. One particularly interesting opportunity presented itself when our tool caught the eye of the organizers of the annual anti-corruption conference CAII at the Comptroller General’s Office of Peru.
With timing working out perfectly, we headed out with Jasmin Grünbaum from NAOF as speakers at the conference to present the level of Finnish data innovation to an audience of more than 1000 conference participants from over 20 countries. More specifically, Risk Detector was presented as part of a panel discussion on the control mechanisms of public officials.
Present with us were also a senior specialist on data-driven anti-corruption work from the World Bank, as well as a money-laundering expert from Mexico. As with our own ideas on how to take Risk Detector forward, there was particular interest on how the methodologies employed in the tool could be used in other contexts such as money laundering or drug trafficking.
The impact of open data
So what to take away as a data professional from attending an anti-corruption conference on the other side of the world? As it turns out, there was one great talking point that was persistently brought up in nearly every presentation I attended – the impact that open data is making in the world.
In terms of anti-corruption policies and activities, open data was highlighted at the conference as a crucial tool to increase transparency in government, enabling tax-payers to take part in ensuring the efficient use of their money. Besides fostering the opportunity to benefit from the work of citizen data scientists, it should also be recognized that open data also opens up resources within governments for data-related projects.
Having gained a global perspective at the conference, the impressive quality, breadth and depth of the open data sources we have here in Finland should be a point of pride that highlights the working design of the processes and standards that enable the existence of this resource. Say what you will about the dusty bureaucracies with endless rabbit holes of processes, comparing to other countries in terms of open data we are clearly doing something right!
Why you should look into open data
Besides the government and its citizens, there are also opportunities for businesses in exploiting open data sources. I’ve listed below two relatively low-hanging fruits that many companies involved with digitalization may find worth looking into.
Risk Detector is at this stage based 100% on open data sources. Working closely with auditors as end-users, our service design methodologies (shout-out to Katariina!) allowed us to develop a fully functional machine learning framework and a demo to answer business questions around a certain area.
While the initial hackathon project was very cost-efficient to the client, the benefit on our side is that the framework that was developed has great potential for generalization. Effectively this means that at this point it is a near-trivial amount of effort to pivot to analogous business questions, and thanks to that we are now able to offer a unique approach to the market.
Besides providing a data source for machine learning models for business, it’s also good for practitioners to recognize that they may often be a good resource just for practicing different methods as well.
Moreover, open data could also be utilized in advanced analytics. Concerning low-hanging fruits, I would point out the potential especially in geospatial analytics: Demographic statistics, for example, are relatively easy to join to geospatial data. This allows analysts to better answer questions concerning demand or behavioral patterns. In another example, open satellite imagery could be used in more specific cases to automate monitoring or progress reporting, for example.
In this article I’ve listed some of the potential selling points for open data. The motivation for this text could of course be seen as a promotion of our capabilities in demonstrably extracting value from data resources, but there is actually another point to be made. The simple reason that I chose to write extensively about this topic is that as open data resources gain popularity, it encourages other parties to offer up their data to the public as well – effectively resulting in a positive feedback loop. The more we use open data sources, the more likely we are to see even better resources in the future.
Happy hacking everyone,