Virtually all major vote-forecasters put Hillary Clinton’s chances of winning in the range of 70 to 99 percent. Everyone from Nate Silver to The New York Times to CNN predicted a Trump loss—and by sizable margins. Even Trump campaign reporter Maggie Haberman said in the Times: “The tools that we would normally use to help us assess what happened failed”. In other words, according to all forecasts, Trump’s election was not supposed to happen. …. but it did happen! 

However, his election did not come as a surprise to us!

The results of our Sentiment Analysis with NABU Socials, which reflects our deep understanding of Social Media, had been telling us that result since the very beginning. We were also told that the most crucial States were in favor of Trump. We did offer free access to our tool, but very few used it and even fewer paid attention to its indications. 

Since the introduction of Internet we have been exposed to major disruptive changes in all the paradigms with grew with (at least those over thirty). It was 1941 when Orson Welles produced the famous film Citizen Kane (a character based in part upon the American newspaper magnate William Randolph Hearst) addressing the issue of media being the Fourth Power (or Fourth Estate).  

That power no longer exists, at least no longer as powerful as it used to be. In addition, the business model has also changed: competition is no longer limited to traditional competitors, but it’s now coming from non-traditional journalistic media as well as from individuals that the social community elects as reliable source of information and opinions.

Trump’s victory has surfaced a latent issue: traditional approaches to market modeling and monitoring are obsolete. The 2016 US Presidential Elections have officially “declared the opening” of a New Marketing Era and its market, The Socials Market, has developed quite rapidly in the last few years. 

A market that communicates via social media where fake news  circulate as fast (if not faster) than true ones. 

It all started with a few adepts, but quickly spread out amongst the youngsters which, as they have grown “older”, have become society and their share of this market is rapidly growing dominant. 

As a matter of fact, we all have changed: to a higher or lesser degree internet and the Apps (Facebook Twitter, etc.) have affected our habits and behaviors. Save for the youngest ones, which are born with them!

Understanding this new market is now a top priority in all businesses and contexts and the questions are: how to understand it, how to reach it, how to communicate to it, or influence it, and how to defend from attacks in such uncontrollable environment. 

Which lesson can be learnt from this experience and how can we build on it?

This document provides a brief description of our suggested “Big Data Approach” to this issue and describes the IT tool that we have developed to assess and monitor the “Sentiment” for the 2016 US Presidential Elections.




No questions: the Socials Market is a product of the evolution in technology and has grown into a very large “Big Data” environment. These two very trivial remarks suggest that any “business” solution addressing this market must be based on a “Big Data Approach” and supported by adequate IT technologies. 

Where to start from? From understanding how this New Market communicates and information travel throughout it. Understand its structure, its Grid. Once the “Socials Grid” is known, business solutions can be developed and their effectiveness monitored.



The structure of the Socials Market is similar to traditional Power Grids, but, unlike them, it’s much more variable, and precisely predicting how info flows through it is much more difficult. However, “difficult” doesn’t mean “impossible”, but only that we need to develop accurate models describing its characteristics (size, behaviors, opinion, etc.).

The best way to understand this market is to “follow” how information propagates through it.

It starts as an input (Level 0) that goes to the addresses I’m sending it to (I’m engaging them). These members of my Engagement (Level 1) may either decide to keep it for themselves or forward it to (or share it with) their contacts (friends) according to their own specific engagement characteristics. And this process goes on and on, until eventually stops. 

The Socials Grid (of my Socials Market) is theoretically designed by this process. As per many theories, its application has several limits which must be accounted for. One above all, the grid’s structure is dependent on the topic being propagated.

In addition the model needs continuous updating as result of both the experience accrued during operations and the evolution in the Socials Market.

However, we can model it and reach an excellent qualitative understanding of its structure and especially of the information propagation pattern. This, of course, requires the development of a sophisticated IT Model!

The most important elements of the grid to be identified are the so-called “Influencers” (red dots on the grid) and the BOT (short for robot) generated posts. 

The Influencers are “users” which have very high engagement levels and also benefit from a good reputation. In other words, they have a relevant impact both on the dissemination of the information and on the “sentiment” attached to it (the opinion on the info being propagated). Identifying them is therefore the most important step in modeling the grid. As a matter of fact monitoring and understanding their opinions and behaviors are key for any marketing strategies.

BOTs are tools used to spread messages throughout the grid and influence it. However, when assessing the market characteristics, they must be filtered out but understood and managed.

Once the grid has been modeled and its relevant nodal elements identified, we can use it to develop Marketing Strategies (the Business Solutions).



Having designed the grid model implies that we also have created an IT Model that can now be used to support the development of (marketing) business solutions. They are necessarily similar to the traditional ones, save that they require the utilization of the IT Model for their implementation. This approach implies four main activities:

  • Understand the Market. This activity is performed both as the first step of any strategy and to monitor the results of any marketing action. The availability of the model makes it possible to scan the Socials Grid and extract all desired information: comments, opinions, expectations, sentiments, profiles, etc.;
  • Develop a Market Reach: The Knowledge of how info flows through the grid and of which are the nodal points (influencers) that need to be reached helps developing communication strategies. Once they have been defined, the IT technology helps the dissemination process through the grid as well as the monitoring (feedback) of its effects;
  • Influence the Market.  The development of a market reach is mainly based on understanding how a specific message would propagate through the grid. As we know, this process is highly affected by the influencers. Having identified and profiled them helps understand how to “influence the influencers”. Once more the IT Model supports the influencing process and the monitoring of its results;
  • Defend from attacks. Fake and true news crowd the Net. The 2016 US Elections has also taught us how disinformation can infect the Socials Grid and influence it. We have also discovered how fake news (the so-called post-truth) are a profitable business. In other words, early detection of any negative message is mandatory to mitigate its potential damages. 

Even in this case the IT model can help both identify the threat (contents, origin, market reach, virality level, etc.) and monitor the actual effects of the counter measures that have been activated.

As already anticipated, there is a fifth activity: exploit the accrued experience to upgrade both the Socials Grid structure and the IT Model. 



In order to provide an actual example of this approach to the Socials Market and a proof of the sophistication of the analysis made possible by our technology, we have chosen to assess and monitor the “Sentiment” on the two candidates of the 2016 US Presidential Election.  This analysis is available at the following link: http://elections.bdt.systems/

We are quite convinced that ours is one of the most (if not the most!) sophisticated, thorough and deep analysis ever done on this subject, mainly because of the potential of our technology.

A few notes on the sophistication of this Analysis:

  • A Real Time assessment of the Sentiment (during the election period)
  • Almost 36 millions posts monitored. A Big Data context!!!
  • Multi-language Analysis, we have chosen to limit our search to the five most important ones (Italian, English, French, German, Spanish and Portuguese)
  • A wide range of Sentiment Analysis possible. As proof of the sophistication of our technology we have designed a tool that allows for a wide range of possible analysis, such as:  
    • Time frame selection. While the Sentiment evolves with time and time-charts show this evolution, it is often necessary to determine its overall values (total and average) over given periods of time.
    • Sentiment per topic. This option makes it possible to breakdown the sentiment analysis per specific topic. In addition, it also shows which other topics (as a %) were quoted with the chosen one
    • Geolocalization
    • Human & BOT selection. This option makes it possible to filter real sentiments form those BOT generated
    • Sentiment of “Originals”. This is the sentiment of the “original” posts, before they are “multiplied” by the engagements.





Sentiments are the result of a never-ending clash between the irrational and rational components of our personality. 

Many variables concur to the end result of this clash: the topic and how relevant it is for us, our knowledge and competence on the subject, our current general mood and also the influence of the external environment (social media above all!).

Our sentiments are therefore the result of a very complex process, as it is assessing them...with a software! However, the so-called Artificial Intelligence (AI) does provide us with some understanding of them.

While ours is a very sophisticated and advanced Big Data technology, it’s up to us to make a good use of its output. This application of our Big Data technology, which is only one of those possible, converts sentiments into numbers with a many-digits precision! However, don’t let you be confused by this mathematical precision. Don’t jump to conclusions. Use your rational component to understand those numbers and integrate them with your experience and good judgment. This is the only way our powerful tool can help you gain a better understanding of...what’s going on!

We hope our tool will help you better understand the evolution of this very important campaign and, above all, the complexity of the voters’ behavior. 

We would also appreciate your feedback on our tool.

Thanks for your  attention and...enjoy playing with our “Sentiment Analysis” 

Roberto Fantino

Founder and CEO - Big Data Technologies srl