Projects and Collaborations

AI Forensics (ex-Tracking Exposed)

Researcher at Tracking Exposed (now AI Forensics, October 2021 – December 2022)
Activities performed: Tracking Exposed (now ‘AI Forensics’) is an NGO founded in 2016 that develops open-source software for the purposes of ‘algorithmic auditing’ and investigates how the recommendation processes of major digital platforms operate. During my time at this organization, I participated as a member of the research group in the design, data collection, analysis, and report writing phases of the following projects:

  • The TikTok Observatory: Interconnecting Junkipedia and Tracking Exposed’s Monitoring Infrastructures (July-December 2022) – This project focused on creating an open-source infrastructure to monitor the algorithmic mechanisms of censorship, content moderation, and ‘shadowbanning’ on the TikTok platform. Specifically, I was responsible for creating a replicable and scalable research protocol for auditing TikTok’s algorithm, which took into account, among other things, the infrastructural differences between the web and mobile versions of the platform.
  • TikTok and the Russo-Ukrainian War (March-August 2022) – This project monitored the evolution of TikTok’s algorithmic policies following the outbreak of the Russo-Ukrainian war, conducted through experimental approaches (‘sock puppets’) and custom software.
  • French Elections on TikTok and YouTube (March-July 2022) – This project monitored the use of TikTok and YouTube as algorithmic tools for political propaganda during the French presidential elections (held in April 2022), utilizing digital methods and comparative quantitative analysis.
  • The Platformization of Gender and Sexual Identities on PornHub (2021) – This project was conducted in collaboration between Tracking Exposed and the University of Milan. The study developed an algorithmic auditing methodology using a series ofsynthetic profiles (‘sock puppets’) and a custom script to understand how the PornHub platform uses its content
    recommendation system to personalize user experiences, processing data related
    to gender, sexual orientation, and geographical origin.
scomodo speranze roma

Scomodo - Le speranze di Roma

Collaboration to the ‘Le Speranze di Roma’ investigation for the Journal Scomodo (December 2020-June 2021)
Activities performed: I collaborated on the ethnographic investigation “Le Speranze di Roma,” conducted by the journal  Scomodo and funded by Enciclopedia Treccani. Specifically, I worked with Dr. Alessandro Gerosa to produce a digital ethnography of the neighbourhoods around the Grande Raccordo Anulare (GRA) in Rome on Instagram, using a variety of methodological approaches (data collection through APIs, hashtag analysis, content analysis, and qualitative visual  analysis).

Dank Memes 2020

2020: Task organizer for Evaluation of NLP and Speech Tool for Italian 2020 (Evalita 2020).
Organizer of the task “multimoDal Artefacts recogNition Knowledge for MEMES” (DANKMEMES). Data collection on Instagram and Twitter, data indexing and manual coding, structuring and evaluation of three tasks based on the computational analysis of the memes’ visual and textual content.

Pomlab

Member of the Public Opinion & Media Lab (POMLAB) Research Group (2022-present)
University of Milan.
Activities performed: Participation in research group meetings, collaboration in developing question batteries, and creating ‘meme-like images’ for the questionnaires of the project “Activating Social and Political Identities Through Political Communication Memes.”