If data mining has a use in the region it is not to do with knife-edge elections
It is a secret to no one. Our smallest online activity leaves a record somewhere. This is called data, which, when gathered, becomes big data. Most online activity records are held by multibillion-dollar companies such as Google, Amazon, Twitter and Facebook, sometimes with users’ unconscious consent.
Many big data companies use a process called data mining to uncover patterns that can help understand society, as well as to change it. It’s alleged that this is what happened in the 2016 presidential election, with Donald Trump’s campaign using information yielded by data mining to reach voters via hyper-targeted messaging. Private strategic communication company Cambridge Analytica provided the campaign the information.
It wouldn’t be quite so easy to do the same in the Middle East and North Africa. In the region, data mining is a more complex process because of linguistic diversity. Information may be in a mixture of dialects. It may be written in Arabic script or in Arabizi, the mix of Latin script and Arabic numerals that is used to communicate in Arabic online or on mobile phones. It could be in modern standard Arabic or Arabic mixed in with some English and French.
This makes it harder to extract and structure information, i.e. to mine data to see clear patterns. Even so, researchers have developed language structuring tools such as Birzeit University’s Arabic Ontology and New York University Abu Dhabi’s Madamira.
The Arab region is one of the world’s most active on social media with 11 million Twitter users and 156 million Facebook members. If their data are mined, it is primarily to do with “sentiment analysis,” also known as “opinion mining,” which has the potential to create an actionable picture for a society based on trends.
People have become more reliant on social networks to express their views on issues that affect their daily lives. In “The Seventh Arab Social Media Report,” put out by the Mohamed Bin Rashid Al Maktoum Global Initiatives, 58% of users express sentiments about government policies on social media.
While this provides an opportunity for authorities to review their policies, examine their effects and people’s view of them, it doesn’t seem to have much political effect.
“The Seventh Arab Social Media Report” also said that 32.4% of respondents were not concerned about using social media and leaving a data trail, which could later be used by governments. For these respondents, a beneficial outcome outweighed other considerations.
Meanwhile, 34.7% said they would continue to use social media but they confessed to having concerns about data mining.
Interestingly, people’s concerns about data mining varied by the sector that it might affect. If data mining were to improve public health care, education and transportation, 38%, 36% and 35% of respondents, respectively, were all for it and not particularly bothered.
An interesting example of data mining in the region was presented at the Third International Conference on Arabic Computational Linguistics in November in Dubai. The Saudi General Authority for Statistics explained that it was trying to understand how public opinion was formed in the kingdom with respect to unemployment. The overall aim is to come up with public policies that mirror public opinion.
In a way, data mining could be an opportunity for the Arab world to re-examine power structures, assess the balance of power among various stakeholders of society, evaluate public policy and leverage data to make society more economically inclusive. However, a crucial factor — the will to do so — is missing.
Instead, there are whispers that data mining has provided crucial information for a recently concluded election in the region. That is unverifiable, of course, and some might ask: Why would there be a need to swing an election which has a result that is not in doubt?
Clearly, if data mining has a use in the region it is not to do with knife-edge elections.