Exclusive content
–
The note you are trying to access is exclusively for subscribers
Subscribe me
Know our plans
and enjoy El País without limits.
–
Pay in
If you are already a subscriber you can
enter with your username and password.
–
–
There is talk of Twitter like the social network of hate. And a quick read is enough to see that what appears to be negative feelings abound: anger, negative criticism, rejection. But is there a place for gratitude, congratulations, or companionship? Felipe Pelzel, director of WILD Fi, and Nicolás Fornasari, data scientist at Alphalabs, created a “social thermometer” that allows us to understand the state of mind of the conversations of a particular community.
And, before breaking it down, we have to give good news: at least in the last six months – a test period for the first version of the system – the indicator registered “a higher percentage of positive feelings”.
The platform that was trained using various algorithms of machine learning that analyze the text of each tweet and determine if their implicit sentiment is positive, negative or neutral, determined the three most positive or “happy” days and the four most “negative” or “sad” days of the period studied.
Let’s start with the bad guys. The day that concentrated more tweets with “negative feelings” in the text, use of hashtags and keywords was on September 9, a phenomenon that spread in the following. It was when it became known that the Senate was postponing the vote for the violation of Guido Manini Ríos.
The tool includes interactive and real-time visualization of community detection. In other words, the algorithm groups the interactions on the same topic within a certain period and it is possible to identify the users who are leading the conversation, either positively or negatively.
In this case, the WILD Fi and Alphalabs social thermometer represented two large, fully polarized communities much larger than the one analyzed by other political or social events.
Another negative day on Twitter was October 20. The spirit was caused by the elimination of Peñarol from Conmebol Libertadores 2020. It was also on November 2 due to the incidents that occurred between people who were in the Líber Seregni square and police officers. Another negatively charged day was January 6, 2021 when the country recorded a record of active COVID-19 cases.
It is striking that the death of former president Tabaré Vázquez on December 6 has not entered the list of the main days that stained Twitter with a negative state of mind that, according to Fornasari, include demonstrations of “sadness, anger or criticism ”.
This is how he explained it: “We could think a priori that the day of Tabaré’s death was going to be a day full of negative feelings but there were so many messages of support and favor that there was not a very strong movement of the balance.
Among the days that Twitter was more positive than ever in the last six months were Christmas and New Years. It was followed by Journalist’s Day on October 23 and the finale of the television show GOT Talent on December 8.
Among the days that Twitter was more positive than ever in the last six months were Christmas and New Years. It was followed by the Day of the Journalist on October 23 and the final of the television program GOT Talent December 8th. “The hashtags relating to GOT Talent they were the most mentioned in the six months, “said Fornasari.
This information can be used, in Pelzel’s opinion, by three main sectors: the public or government, companies in general and advertising agencies. The goal is to “understand people better” and, for example, this is useful when communicating or launching campaigns.
El País and the company IGV also apply artificial intelligence to analyze thousands of public posts on Twitter. This is how information has been gathered during the health emergency and the last electoral days. The classification of the tweets is analyzed with these criteria: positive (hope, humor, encouragement, solidarity), negative (impotence, anguish, fear, complaints) and neutral (an informative tweet, for example).
– .
