This project, Online-SoBA (Online Social Behavior Analysis), is a semi-automatic system capable of analyzing short texts corresponding to comments written in Portuguese as a reaction to 'posts' on social media platforms (namely social networks or online newspapers), so that it is possible to identify behaviors that define social opinions at a given time.

To face this challenge, the texts contained in the NetLang corpus are being analyzed, and information regarding the theme and polarity is extracted for each one.
The calculation of the polarity of a comment is carried out through the number of words with positive or negative connotation that constitute it. Subsequently, according to sentence structures that characterize these behaviors in natural language, the theme of the comment is assigned.

By analyzing the resulting graphs of this process, we are quickly able to make a visual deduction about how social opinion on a topic has evolved over time.