Social media and COVID-19

Even though there are databases of tagged digital content for different purposes, there are few resources that are representative of Colombian Spanish present on social networks. Likewise, transfer learning (i.e., using models trained in other languages) using transformers and multilingual numerical representations or implementing translation-Re-translation strategies are not enough to capture the richness and detail of this geographic variant of Spanish. Because of this, we decided to design and tag a database of tweets and posts on Facebook collected since the beginning of the pandemic around the following themes defined by COLEV: 1) vaccination, 2) education and schools, 3) domestic violence and 4) mental health, with an emphasis on identifying vulnerable populations. This database will serve as input for automatic content classification algorithms.

The topics that are discussed in social networks change over time and depend on the social, political, health, and economic phenomena that occur in parallel at different geographical scales. The objective of this board is to visualize the dynamism of the topics related to COVID-19 that have been discussed on Twitter and Facebook since the first COVID-19 case was reported in Colombia, along with their main engagement metrics, such as likes, comments and retweets / shares, among others.

In order to determine the scope, relevance and representativeness of the analysis of social networks in Colombia, we made a description of the penetration and use of internet and social networks in Colombia based on: 1) Data from the 2020 Americas Barometer (LAPOP) , 2) the Global Digital Yearbook (2020) by We Are Social and Hootsuite, and 3) studies carried out by the Ministry of Information and Communication Technologies of Colombia.