In January 2015, terrorists shot and killed 17 people in and around Paris, in the offices of the satirical magazine Charlie Hebdo, at a kosher supermarket a few miles away, and in the southern suburb of Montrouge. These attacks, the deadliest in France in more than 50 years, prompted vigorous—and still ongoing—debate over issues ranging from free speech and satire to immigration, racism, and religion.
From the outside, it seemed that, socially and politically, France was essentially being split into two camps: those citizens who looked forward with optimism, hoping to defend the values that had made France the target of the jihadis, and those who were caught up in the cycle of resentment and revenge.
We at SAP Startup Focus wondered whether we could analyze all the sentiments people were expressing online in both traditional and social media channels, to make a true, data-based assessment that might answer a simple question: After Charlie Hebdo, did the level of consensus about the value of multiculturalism and religious freedom grow in France? Or did the country risk being divided by the politics of fear?
Semantic Visions, a Czech Republic–based company that has the world’s largest semantic news database, recently conducted just such an analysis. Its researchers identified the terms most frequently used by people expressing their feelings about the Paris attacks. Then they tracked those terms across multiple media and in multiple languages both before and after the attacks. The results of that analysis might be described, without too much hyberbole, as a snapshot of the collective pulse of humanity on these issues.
Semantic Visions found that consensus-model terms outnumbered conflict-model terms both before and after the Paris attacks, across all the media languages studied.
Semantic Vision’s Models
Following are specifics about the methodology Semantic Visions used to track the worldwide online expression of these sentiments:
Semantic Visions tracked the usage of these terms for four weeks before the attacks and four weeks after them. The research involved 268,000 separate media sources and examined nearly 49 million individual articles in multiple languages. The researchers filtered these results to include only those in French, English, German, and Czech.
For the consensus model, they searched for all words, phrases, and associated synonyms in all possible linguistic forms for concepts general viewed as positive: democracy, dialogue, free speech, love, multiculturalism, peace, tolerance, and unity.
Similarly, for the conflict model, they tracked the following terms, which sometimes carry negative connotations: clash of civilizations, confrontation, cultural conflict, hatred, intolerance, Islamophobia, revenge, sharia (Muslim law), and war.
In conducting its research, Semantic Visions relied on two key assumptions:
In democratic countries, the media are usually free from government control and individuals are generally unafraid to express their genuine opinions on line.
The two-month timeline (one month before the attacks, one month after) was long enough to accurately represent the baseline public opinion before the events and the reaction following them.
Results: Consensus Trumps Conflict
Overall, Semantic Visions determined that consensus-model terms outnumbered conflict-model terms both before and after the Paris attacks, across all the media languages studied.
In addition, researchers found that use of conflict-model terms after the attacks decreased faster than the use of consensus-model terms, meaning that that the social consensuspersisted while the social conflict diminished. This significant finding provides hope for the future, and ought to serve as a reminder for French politicians as to what their constituents are really thinking.
Source: Semantic Visions. Used with permission.
Not unexpectedly, where issues of violence, religion, and race are concerned, even the “public pulse” is seldom a monolithic entity and several exceptions can be observed, as the findings indicate:
The highest growth rate in usage of terminology detected in French-language publications was for these two consensus-model terms: free speech (439%) and tolerance (160%). The highest growth in the French-language conflict-model topics included: cultural conflict (514%), sharia(208%) and Islamophobia (181%).
The usage of the term antagonism decreased very slowly. Even one month after the attacks, it was still 30 percent higher than it had been in December 2014, before the attacks. On the other hand, the term dialogue exhibited a slow but persistent upward trend, potentially indicating that, eventually, society will recognize the need for intercultural or interreligious reconciliation.
Terms that got a short-term boost (because of the relentless news cycle), but which retreated to baseline levels within a month, included cultural conflict, free speech, andmulticulturalism. At the same time, some of the terms that didn’t retreat to prior levels included the terms hatred,intolerance, Islamophobia, and revenge.
Big Data, Big Insights
What this study indicates is that, no matter what the talking heads on TV are saying, and no matter what radio, cable, or Internet channels you follow, the reality of public thinking is always somewhat more complex than what we see or hear in the media. Today’s technology—more specifically, big-data tools—can provide incredibly valuable insights into the public mindset and the real sentiments that citizens are expressing whenever such critical issues come to the fore. For more insights, please read the Semantic Visions report in its entirety.