A group of researchers from the University of Granada (UGR) has devised a system based on artificial intelligence that predicts election results, just by analyzing Twitter posts.
Tweets? Yes, as you read, scientists propose an AI capable of handling large amounts of information, massive, well, this is a very wide network, where an unlimited amount of content is shared.
To deal with the “unstructured information” that is shared on the platform, the researchers propose a descriptive Big Data system.
What is a descriptive Big Data system?
It is a process that consists of analyzing an immense amount of information more easily. The system groups data that share similar characteristics, and that in turn are easy to manage.
The purpose of this technique is to summarize what has happened to a certain event, and then explain it in simple words.
In this case, the University of Granada has made use of this system to analyze the behavior of Twitter users, in relation to electoral elections and to forecast the results thereof.
This system was used in the last presidential elections of the United States to predict who would win the contest between Donald Trump and Hillary Clinton. We already know who won.
Currently, Twitter is one of the most important social networks. On this platform, as we already mentioned, we find a dense amount of information. It can even be said that it is the favorite network of politicians. For example, Trump uses it very often to interact with his followers.
But how does the intelligent system devised by the UGR researchers work?
Basically, the system seeks to group tweets loaded with emotions and feelings, both against and in favor of the candidates. In this regard, the research team expressed:
“Today, these techniques are of enormous value because they provide easily interpretable and easily understood solutions. They allow easy data traceability and provide easily explainable results that can be used by people without technical knowledge, thus democratizing access to artificial intelligence ”.
What are the emotions that the UGR lA takes into account? they can be both positive and negative feelings.
The artificial intelligence presented by the University of Granada detected a higher percentage of annoyance “towards Hillary Clinton than towards Trump” in the last presidential elections. Why? We do not know.
On the other hand, he perceived excess confidence in the tweets shared by Trump and his followers, regarding his possible victory, which in the end became a reality.
According to TechXplore, other patterns used by the system may be related to stronger words, allusive to “ban, service, transgender and Donald Trump.” So, to look for associations between these, and the possible emotions that it caused at the time, or that may continue to cause.
Now, the big question that remains is, will this team of researchers already have a forecast of the recent presidential elections in the United States?