Big Ideas for Big Data in Belfast

A review of the Big Data Belfast 2019 conference #BigDataBelfast19 by Barbara Franzoni.

Now in its 7th year, the Big Data Belfast 2019 conference took place in the beautiful conference centre that is ICC Waterfront. As far as we know, this is the biggest data science conference in the UK to date. On arrival, delegates were greeted with an agenda, a goodie bag and a warm Belfast smile.

Then it was into the exhibition hall for a chance to mingle with the many exhibitors over a cup of coffee, where I was delighted to find that all were more than happy to talk with me about ideas, innovations and the many possible applications for big data and data analysis. Soon the exhibition hall filled with the 600 delegates who were registered to attend the event but the Waterfront is so spacious that it did not feel overly crowded.

Opening the Conference

In his opening address, Roger Woods commended the tech, AI and data science communities for making Northern Ireland, and particularly Belfast, one of the best UK tech locations outside of London. He referenced several reports including the Matrix Report and Tech Nation, which show data science, AI and other examples of where Northern Ireland is well ranked. Roger also mentioned the growing job creation in Northern Ireland referencing two huge funding initiatives; Techstart and Investor Present Capital. 

Dr Aislinn Rice of event organisers, Analytics Engines, told us how proud she was to be part of an inspirational team in a homegrown company. She invited attendees to speak with Analytics Engines team of problem solvers to “Challenge the Expert”. One of her comments I enjoyed was that we are all telling stories with data, but data is most powerful in the hands of problem solvers who see the world differently. 

Following a pretty phenomenal opening video, the big data speakers graced the stage and while all were extremely noteworthy, I’ve listed below just a few of the moments that I found most memorable.

9 Notable Shifts in Data Science

Eoin O’Reilly, Head of Analytics at EY Ireland (chief sponsor for the event) gave a very enjoyable look back at the past decade of data science and listing nine significant shifts he has noticed, as follows:  

1. The evolution from number-crunching to storytelling that has resulted in different skillsets being required in the industry.

2. Data Science now has a seat at the executive table 

3. Data for Good. The over-riding aim in data science is to make life better for people, e.g., optimising journeys, managing patient flow in hospitals, working with transport authorities to make bus networks more efficient. 

4. Rise of the Robots. Data analytics projects once sought outcomes to drive decisions, but people now use the information received from Alexa and Google Home to make decisions instantly. These technologies use smart analytics, which is under the hood of core platforms.

5. Data for Bad. We’ve seen serious allegations of misuse of data interfering in elections and impinging on core democratic values. As a society, we’re starting to grapple with these risks and we should see interesting developments in this regard over the next few years, including regulations that will go beyond GDPR to protect citizens. How we develop algorithms and deal with inherent bias in data sets is also important as bias leads to poor outcomes. Here Eoin cited the notorious example of attempts to predict reoffenders in US prison systems.

6. Money Ball Effect. Sport and sports analytics have been great for data science and Eoin cited NI company, STATSports as a great example of how data can help team performance.

7. Data Democracy. We have so much more tech at our fingertips that is available and accessible for all. Data and analytics are no longer privy to the select few.

8. Levelling the playing field. While gender diversity is improving, which addresses many of the inherent bias in data, more work is needed to ensure true diversity in gender and other areas. 

9. Minds and Machines. There’s still a big human element in analytics and we’ve come a long way in how we think about and frame problems as well as post analytics change management. 

Talking Evolution and Revolution

After the first panel in which the benefits of big data and locating in Northern Ireland were discussed, Dr Clair Sullivan of GitHub spoke about data evolution, referring to the cloud as being just other people’s computers and pointing out that data analysis actually dates back over 7000 years as Mesopotamian records exist showing data being used to analyse and predict crop cycles and farming conditions. 

Is Higher Education as relevant as it once was?

Clair spoke about how the field of data science appears to be reactionary but evolution is what we should strive for. She then turned to evolutionary learning and hiring. Where once only PhDs were considered for data scientist jobs, we now see more bachelor degree educated doing the same job. Learning is evolutionary because it is continuous and with data science changing continuously, self-learning is necessary to keep up. Therefore formal education may not be as important as it once was, and a continuous learning curve consisting of many short-term learnings may now be more relevant.

Clair summed up by saying that revolutions will come, but it’s more important to carry on evolving. 

After coffee break, I wandered over to the smaller Business Analytics and Digital Transformation stage where I found some great talks on digital transformation in traditional sectors such as construction and even more interestingly, how big data can be drilled down to track and predict arms conflicts through international trade in small dual-use components.

Trade is the fuel of armed conflict

Rebecca Harding, CEO of Coriolis Technologies gave the most fascinating talk on how data science in international trade can be used to track and predict armed conflict. This is done by analysing traffic in dual-use goods, that is, goods that have both military and civilian uses with possible applications such as missile launching components. As an example, Rebecca took us through the anatomy of an AK47 rifle to look at a very small component; a steel U-Section. When strong trade in these u-sections was noted between Russia and Ukraine, it pre-empted hard conflict. Tracking large trades of this good has successfully predicted violent outbreaks globally.

Rebecca went so far as to call “trade” the fuel of conflicts but warned that some countries are more difficult to extrapolate data from than others. She noted that wherever sanctions occur, so too does the reclassification of good, services or even trading partners in order to get around the sanctions. Sanctions, she said are the weapons of choice for politicians that make it hard for data science to track dual-use goods. However, as the adage goes, follow the money to expose the truth. It’s for this reason that banking is the new frontline in tracking armed conflicts.

The post-lunch cybersecurity stage was a scary place as many great talks were given on hacking and viruses. Speakers from Dell, ZScalar, Analytics Engines, Allstate and RBS came with an absolute wealth of detailed advice on preventing, dealing with and recovering from attacks or big data incidents.

GDPR had to happen; Marketers have invaded people’s privacy for years!

Steve Rogers of Reach PLC gave a very interesting (and almost revolutionary) talk on the difference between data targeted advertising and ads that utilise contextual relevance.

In recent times, digital marketing has concentrated on targeting and retargeting ads to ensure relevance to the audience. Relevance is important to the efficient serving of ads to those who are more likely to be interested in your product. However, targeting needs personal information, and retargeting (e.g., the same ad following you around the internet) can be intrusive, invasive and downright creepy.

Steve said that GDPR had to happen, marketers have been invading people’s privacy for years. He also warned that GDPR breaches will be costly. Third-party cookies are starting to disappear too, so the way advertisers can personalise ads becomes less precise – meaning it’s less likely you’ll get the ad for the one product you want at the one time you need it.

Steve then examined the efficiency of contextualisation, explaining the concept with the example of a glossy fashion magazine where ads can make up more than 60% of the content. In these examples, the ads were found to be so relevant to the reader and reflected other content so closely that they were perceived as general content rather than ads. Such magazines only include advertisements where the quality and relevance are consistent with overall magazine content and Steve suggested that this should be done online too, with creative directors checking ads for consistency with the rest of a website (the context).

Steve mentioned that online influencers are better at contextualising ads than most businesses, as they curate their brand by choosing ads from brands that share their values and are equally relevant to their audience. This is a great example of contextual relevance and consistent branding.

Reach has been running some experiments on contextually relevant ads and the results (as illustrated in the graph tweeted above) show that contextualised relevance is not only less invasive than personalised online advertising, it’s also more effective.

Still Bubbling with Ideas

There were more talks including the use of data to uncover scandals and ethics in AI and data analytics. Finally, the day finished with a fascinating panel discussion on the use of data in sports including tracking of performance and player wellbeing after injury. 

Our hosts, Analytics Engines, then invited all delegates and speakers for some closing drinks where there was much discussion of all the ideas and possibilities that the day had highlighted. All in all, it was a fantastic and mind-opening event that I’m sure will result in more possibilities explored and ideas hatched – some of which will perhaps be launched at Big Data 2020.