EPIC Lab: Effective Programming, Interaction, and Computation with Data

A group of UC Berkeley researchers – including BIDS Faculty Affiliates Aditya ParameswaranSarah Chasins and Joseph Hellerstein (Berkeley EECS), Erin Kerrison (Berkeley School of Social Welfare), and Senior Research Data Scientist Stéfan van der Walt – have received a 3-year, $2 million National Science Foundation (NSF) grant to improve the useability of big datasets and create new computing tools that will help defense attorneys, criminal justice workers, and others more easily engage in data-intensive research into police misconduct, judicial decision-making and related issues. The Berkeley Institute for Data Science (BIDS) is a partner, and other collaborators on the project include Niloufar Salehi (Berkeley School of Information), and Berkeley EECS’ Joseph Gonzalez, Anthony Joseph, and Koushik Sen. 

The new Effective Programming, Interaction, and Computation with Data (EPIC) Lab, is also collaborating with the National Association of Criminal Defense Lawyers (NACDL), Secure Justice, and the San Francisco Public Defender's Office, groups with considerable criminal justice expertise that will be contributing to help shape the goals for the tools the lab is developing. 

The tools being developed will help defense attorneys use and understand large quantities of data about policing, courts, and individual cases – information that is currently locked up in isolated, disparate reservoirs of messy and incomplete data, scattered across scans of paper forms or hours of audio and video records. The project's data platform will leverage and advance three key underlying techniques, including (1) no-code and low-code modalities like natural language search boxes and spreadsheet interfaces, (2) program synthesis and machine learning to transform "fuzzy" queries into a space of possible interpretation, and (3) interactive ambiguity-resolution widgets that present visual representations of output data. The development of this platform will contribute to advances in program synthesis and ML-aided program generation, and these techniques will be incorporated into a larger user-centered design process working toward tools and interfaces that meet public defenders’ needs and take into account the legal context and constraints in which they work.

The EPIC Lab team will partner with legal associations and defense attorneys to develop data analysis methods and tools that faciliate the work of collecting, organizing, and suggesting analyses of vast public and private datasets. Addressing this significant gap will increase the efficiency of defense attorneys and their staffs, increase the value of the data, and ultimately lead to fairer, better outcomes in criminal justice contexts. Altogether, the work will contribute to building more efficient and usable no-code and low-code tools to democratize data access more broadly.

The tools being developed will initially be used in San Francisco, Alameda, and Sacramento, and one of the datasets that EPIC researchers will work with is a large set of California police records, released in response to a 2019 state law, that enable defenders to search for state police officers with records of lying, withholding evidence and other forms of misconduct. The collaboration will also contribute to a pre-existing collaboration between UC Berkeley and the criminal defense lawyers association known as the Community Law Enforcement Accountability Network, a project that aims to expand a Legal Aid Society database in support of public defenders in New York City.