Professor and Researcher
DaVint Lab member
DaVint Lab was created in 2017 by the PUCRS professors Milene Silveira, Isabel Manssour, and Roberto Tietzmann. It is a mature version of previous initiatives from the researchers, that have been working with data visualization since 2011, with an interdisciplinary team (joining the schools of technology and communication) It is physically located in Room 654 of Building 32 of PUCRS. I was invited to collaborate in 2018, and I am still a member to this day.
Data visualization, visual analytics, computer vision, interactive digital narratives, and end-user development are the main areas of research of the projects under development in the lab. You can see some of our works below!
Humor, support and criticism: a taxonomy for discourse analysis about political crisis on Twitter (2018)
Authors: Carlos Roberto G. Teixeira, Gabriela Kurtz, Lorenzo P. Leuck, Roberto Tietzmann, Daniele R. de Souza, João Marcelo F. Lerina, Isabel H. Manssour, Milene S. Silveira.
Objective: to propose a taxonomy for discourse analysis to help understand the intent and types of Twitter users’ conversation about political scandals.
Context: The use of social networks to share information and express opinions has significantly grown in recent years. In this context, the microblogging platform Twitter has been used in the political scene by governments and citizens for different purposes. Brazil is experiencing a political crisis in recent years that led to the impeachment of Dilma Rousseff president in 2016 and the investigation of several corruption scandals in 2017.
Contribuitions: In this paper, this taxonomy is described, and an analysis is presented about four political episodes that occurred in Brazil in 2017. It was possible to verify how this taxonomy can help to identify different types of user manifestations, allowing us to analyze patterns of behavior. The main contributions include: providing a new taxonomy to enable the opinion analysis about political facts and crisis; investigating the perception of social media users in Brazil about political points; and identifying the benefits of using visualizations in this context.
What was my role in this research?
I developed the methodology, based on Content Analysis (Bardin and Krippendorf), to analyze and categorize the datasets. We chose 4 case studies that featured political scandals in Brazil from May to September of 2017, and gathered tweets made by users. Since we had millions of tweets, I proposed to use the 100 most retweeted. Then the team brainstormed the categories by reading them.
Once we settled with 5 (news, support, humor, criticism and protest, and neutral), we did peer review, where two different researchers qualitatively classified each tweet, and in the case of a tie, a third evaluator from the team determined the final category of the message. In the image below, you can see the percentage of the categories we found in all datasets:
I supervised the process to guarantee the methodology was being correctly used. After organizing the data, we gathered the team to discuss insights and results. I also was responsible to write the metodology section of the written article and to do the final proof-reading.
The telenovela Dona do Pedaço between television and Twitter: convergences and divergences in online repercussions (2020)
Authors: Roberto Tietzmann, Carlos Roberto G. Teixeira, Diego F. Furtado, Gabriela B. Kurtz, Lorenzo P. Leuck.
Objective: to understand how the audiences behave in Twitter regarding TV Shows. We chose the "A Dona do Pedaço" novela's last episode exhibition (november, 2019) and used a took we created called PeakVis to sync video and tweet dataset to understand the behavior simultaneosly.
Context: Soap operas are the main fictional product of Brazilian television, and their ability to engage audiences makes their plots and characters traditionally motivate debates among their viewers. Four out of five viewers watch the programs with the phone as a second screen (THINK WITH GOOGLE, 2018). Thus, it is safe to assume that such debates were also developed through Social Network Sites (SNS).
Contribuitions: We designed and developed TweetUtils, a set of tools for capturing and viewing Twitter posts, and PeakVis, which allows you to synchronize a video or audio record with a dataset collected in SNS, generating interactive visualizations. In this article, we showcase the potentiality of our tools in 2 ways:
(1) detect and analyze tweet volume peaks, relating them to the narrative and audience reactions;
(2) understand the relationship between the time elapsed in the broadcast and the audience's reactions.
You can downoad all the tools we developed at our GitHub, clicking here.
What was my role in this research?
I helped in the development of PeakVis, testing the application and suggesting improvements in usability and UI, from the user's perspective. I also took part in the research writing the methodology procedures and some final considerations.