17th AIAI 2021, 25 - 27 June 2021, Greece

Recognition of epidemic cases in social web texts

Apostolos Antonakakis, Eleftherios Alexiou, Nemanja Jevtic, Georgios Sideras, Eftichia Farmaki, Sofronia Foutsitzi, Katia Lida Kermanidis


  Since December 2019, Covid-19 has been spreading rapidly across the world. Unsurprisingly, conversation in social networks about Covid-19 is increasing as well. The aim of this study is to identify tentative Covid-19 infection cases through social networks and, specifically, on Twitter, using machine learning techniques. Tweets were collected using the data set “Covid-19 Twitter”, between November 1, 2020 and December 30, 2020, and manually marked by the authors of this study as positive (describing a tentative Covid-19 infection case) or negative (pertaining to any other Covid-19 related issue) cases of Covid-19, creating a smaller but more focused dataset. This study was conducted in three phases: a. data collection and data cleaning, b. processing and analysis of tweets by machine learning techniques, and c. evaluation and qualitative/quantitative analysis of the achieved results. The implementation was based on Gradient Boosting Decision Trees, Support Vector Machines (SVM) and Deep Learning algorithms.  

*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.