Project description:

Mina57 Data-Selfies collected data from Twitter-Peru users in order to obtain a map of political position concerning the dissolution of the Peruvian Congress on September 30, 2019. Supported by a Knowledge-Database populated during the last two years, the further analysis of this map, allowed us to measure the effect of social media “Echo Chamber.” We follow the “Code of conduct for professional Data Scientists” .
 

Impact reached:

• First of all, statistics and technical reports in Peru have a small audience. Furthermore, maintaining a neutral position brings upon you all the radicalized accounts attack. 
• We posted on Twitter a summary of the project report. The Twitter post thread obtained 60K impressions, 3K interactions, was recommended by the former INEI (National Institute of Statistics and Informatics) chief, and doubled the quantity of our Twitter account followers. 
• Our conclusions have shown that the “Echo Chamber” effect was increasing polarization in the political discussion. The maps generated displaying interactions between Twitter accounts shown two main groups: “Fujimorismo” and “Anti-Fujimorismo“. Because our approach was without political bias, the report introduced as a project product was received for both communities as an input for further analysis. We received 4K visits on the website where our report was published. 
   

Techniques/technologies used:

• Twitter APIs, scrapping, data cleaning, Python: Used to create a framework that enables the collection of posts published on Twitter, accounts profiling, and networks discovery. 
• Data visualization, Gephi: Used to analyze the networks created in Python. We applied OpenOrd, YifanHu, and ForceAtlas2 algorithms. 

What was the hardest part of this project?

• In order to understand Twitter-Peru communities dynamics, we have built a Knowledge-Database with rules that evaluate a set of variables that change in a significant-political-event basis. We do not identify individual accounts, but those handled by public figures or recognized Twitter activists. We are not applying machine learning because models became obsolete very quickly. At this time, we can apply this tool to update a map populated with Twitter-Peru communities. From a technical point of view, designing and building this Knowledge-Database using only Open Source software made the hardest part of the project.

 

• Peruvian citizens trust neither established media nor government reports. Polarized discussion in social networks favors fake-news propagation and the “Echo-chambers” effect. This project tries to be politically unbiased and provides quantitative information about the quality of Twitter accounts (age, spam rate, activity, popularity) discussing Peruvian politics and the relations established between activists, news providers, and regular accounts. This way, we aim to earn public trust in our data (we provide data tables on our website) and introduce in the discussion the need to listen to everyone, even when the other voice opposes your political position. Therefore, from an effective communication point of view, find a way to better deliver our message was the hardest part of the project.

 

What can others learn from this project?

• The effect of similar-minded groups isolation in political discussion. We need to listen to each other and open our minds to different points of view. 
• Data visualization techniques. We have been approached to teach some lessons related to Gephi use, but we are not able to provide these services. Instead, we have answered the requirement with a road-map, including MOOC courses and Open Source technologies.

Project links:

www.mina57.com/mapa-de-ubicacion-politica-de-activistas-en-twitter-sobre-la-disolucion-del-congreso/

www.mina57.com/en/map-of-political-position-of-twitter-activists-about-the-dissolution-of-congress/

twitter.com/minacincosiete/status/1211077222800535557

twitter.com/matuk/status/1211110733938855936