Current DSSG Projects
Since 2022, the Data Driven Discovery Initiative has provided seed grants through the Data Science for Social Good (DSSG) program. These grants support projects that address wide-scale societal challenges through innovative data science methods and interdisciplinary scholarship. Past projects have addressed questions of climate change, immigration, justice, and public health. Our current DSSG project recipients, listed below, support projects that touch on topics such as threats to democracy, public policy, conservation, and the justice system.
Apply for our DSSG faculty seed grant
Modeling Climate-Induced Societal Adaptation and Population Displacement with New Machine-Coded Environmental Event Data
From Nigeria to Pakistan and Central America to Small Islands States, political conflicts have erupted over resource scarcity exacerbated by climate change. Our project focuses on how dozens of vulnerable developing countries respond and adapt to climate change, and the extent to which those adaptations promote conflict. To date, a huge challenge in understanding climate impacts is the lack of high-resolution, high-frequency data on how people and governments are responding to climate change. To address this shortcoming, we will use a series of fine-tuned Large Language Models (LLMs) applied to a unique corpus of 130 million articles published by local media in 66 developing countries from 2012-2024. This will allow us to create a monthly, sub-national dataset on 10 distinct types of environmental adaptations. Using this data, we will model the relationship between climate change and adaptation at an unprecedented temporal and geographic scale. Our key outcomes will include major societal disruptions studied by extant research, such as population displacement and civil conflict, but also previously unmeasured adaptation behaviors, such as protests, civic activism, and legislative action by governments.
Team: Irina Marinov, Erik Wibbels
AI for Philadelphia: Mapping Illegal Dumping Hotspots
Illegal dumping costs Philadelphia taxpayers millions of dollars annually in cleanup efforts. It lowers quality of life, degrades green spaces and waterways, and disproportionately affects socio-economically vulnerable communities. In partnership with the City of Philadelphia’s Department of Parks & Recreation and the Water Center at Penn, DDDI’s research team is leading an analysis of over 100,000 city service reports to understand and predict patterns of illegal dumping across Philadelphia. By combining advanced spatio-temporal statistical models with transformer-based image classification, the project identifies persistent hotspots and helps city agencies design more effective interventions, with a focus on community parks.
Team: Elena Liang (DDDI Data Scientist and Project Lead), Terhi Nurminen (DDDI Undergraduate Summer Research Assistant), Colin Twomey (DDDI), Natalie Walker (City of Philadelphia Dept. of Parks and Recreation), Jazmin Ricks (Penn Water Center)
Wearable Gunshot Detection for Public Safety Applications
This project will collect high-fidelity measurements of firearm use to support the development of algorithms and sensors designed to reliably detect and classify gunshot events in the context of public safety applications. Many law enforcement agencies struggle to record and reconstruct firearm use by their officers due to technical limitations in the body-worn cameras (BWC). In parallel, many community supervision programs struggle to ensure that prohibited possessors refrain from using firearms during periods of court-ordered community supervision. By acquiring multi-modal sensor data on the gunshot events, this project will address these twin problems by establishing ground-truth measures of signal propagation and degradation.
Team: Charles Loeffler
Machine Learning for Peace
Erik Wibbels, professor of Political Science, has been awarded a DSSG to expand the Machine Learning for Peace (MLP) project. MLP leverages machine learning to provide timely forecasts on civic spaces for policymakers and activists to use to defend democracy in an increasingly repressive world. With the DSSG grant, the project will now include a new predoctoral program aimed at inviting traditionally underrepresented students to become integral members of the project team. These predoctoral students will contribute to expanding the breadth of countries covered by MLP and use high-quality data to improve forecasts for the project. Predoctoral trainees will have opportunities to publish novel research equipping them with skills to become competitive applicants for top PhD programs globally.
Team: Erik Wibbels
Philadelphia City Lab
In Fall of 2024, with support from this DSSG grant, Political Science Professors John Lapinski and Marc Meredith as well as data scientist Samantha Sangenito from the Fels Institute of Government are launching the Philadelphia City Lab. City Lab will offer fellowship opportunities for Penn students studying data science to work directly with Philadelphia city council support staff. These students will conduct their own original analyses on ongoing public projects to support legislative action. This initiative enables Philadelphia City Council members to make more efficient, equitable, and data-driven decisions while offering Penn students valuable hands-on experience in data science roles and preparing them for careers in public service.
Team: John Lapinski, Marc Meredith, and Samantha Sangenito
Black Representations of International Governance, an Extension
Political Science Professor Julia Gray and doctoral student Chloe Ahn, recipients of a 2023 DSSG seed grant, have been awarded an extension of their grant for 2024. Their project aims to understand the historical role of Black newspapers in advocating for issues important to the Black community on an international scale, which has been absent or misrepresented in mainstream media. In their work so far, the team has trained word embedding models to digitized newspapers published between 1940 and 1980. These models identify key differences in language usage between Black and white newspapers. The extension of this seed grant will allow the team to continue rigorous data cleaning and analysis as well as expand the analysis to fullpage files from newspapers, in addition to individual articles.
Team: Julia Gray and Chloe Ahn (PhD student)
The Use of AI in Forensic Science: Accuracy and Fairness of Facial Recognition Technology as Used by Law Enforcement
Criminology Professor Maria Cuellar will be studying the accuracy and fairness of facial recognition technology (FRT) with the 2024 DSSG grant. Given the increasing use of FRT in police departments and its potential for bias, this project aims to assess the accuracy and fairness of FRTs in the kinds of images used by the police. On completion, the project will provide a resource for police departments seeking to minimize the harms done by FRT and change the way this technology is scrutinized before it is employed. Cuellar intends for the research to be a resource for judges, attorneys, and scholars alike in evaluating the impact of FRT in criminal trials.
Team: Maria Cuellar and James To (PhD student)
Forecasting Regime Shifts in Social-Ecological Systems using Universal Differential Equations, with the Amazon Basin as a case study
Post-doctoral researcher Emerson Arehart and professors Erol Akçay and Joshua B. Plotkin of the Department of Biology will address the challenges of modeling and forecasting regime shifts in complex ecosystems negatively affected by human activity. Their project focuses on social-ecological systems, like the Amazon Basin, where human behaviors interact with natural processes in unpredictable ways. Using a new technique called Universal Differential Equations (UDEs), which combines neural networks with traditional mathematical models, they aim to improve predictions of sudden ecological shifts. By capturing the complex dynamics within these environments, this approach could help anticipate and potentially mitigate disruptive changes in vital ecosystems.
Team: Erol Akçay, Joshua B. Plotkin, and Emerson Arehart