BA or MSc projects
All these projects require a solid Data Science background. What does this mean? Please use this guide as a starting point: https://nerds.itu.dk/students/. Depending on the project, more competencies might be required. Please find the project below.
Auditing algorithms: This project has a strong literature/code review component. You will review the approaches to audit algorithms that are based on data, like those that rank products on Amazon or that prioritize notifications about vulnerable children. You will work on reviewing algorithms where (i) we know the input and output data, but not the functioning of the algorithm, which is a black box, (ii) we do not know the input or output data, but we know the algorithm, which is a white box, (iii) we have no access to data and the algorithm is a black box, but we can do reverse engineering of its functioning by interacting with the algorithm. The underlying questions are: are these algorithms biased? Do they do what we expect them to do? In the second part of the project, after reviewing the approaches, you will be in charge of auditing a system/algorithm, using one or more of the reviewed approaches. Requirements: foundations of algorithmic fairness and explainability.
Debiasing careers : Do careers of scientists evolve in an unequal way? Can we provide an unbiased visualization of career success? You create a web application that takes in input publications and citation data for a scientist (e.g. from google scholar or web of science), applies science of science knowledge and fairness measures to debias indicators, and generates a “career” similar to this or these.
Changes in the collaboration network during the pandemic: Science is a collaborative endeavor and the COVID-19 pandemic has disrupted the way we collaborate, and therefore how we do science. But how? How has the collaboration network among scientists changed? Who has been most affected? Has the pandemic exacerbated inequalities among scientists? In this project, you will use large-scale databases about publications and citations and will construct a temporal network of scientific collaborations. You will study the changes that occurred in the period 2020-2022. This project has both programming-data challenges (you will be working with networks with tens of millions of nodes and up to billion of links), and a science of science/social scienc-y component. So a team with diverse skills is required. Requirements: Strong knowledge of network science, strong programming skills, preferably in parallel computing.