New thesis projects at the OpenGov Lab

The OpenGov Lab has six new thesis projects for (Master) students who want to work on AI for open government. The projects range from e-discovery for FOIA requests to how LLM summaries of government documents leave out what matters to citizens. Download the flyer (PDF).

1. NLP/IR for e-discovery: optimize any part of the high-recall pipeline that government legal teams use to find relevant documents for FOIA requests.
2. Helping citizens formulate FOIA requests without exposing sensitive data: build systems that help citizens specify what they are looking for, without leaking personal or other sensitive information.
3. Delete or archive?: test whether BERT models and small, local LLMs can decide which e-mails must be archived under the Archiefwet, and whether their explanations are plausible.
4. Proactive disclosure: match newly published government documents to the citizens who need to know about them.
5. Extracting and consolidating compliance requirements from legal text: use LLMs to extract and consolidate requirements from overlapping standards such as ISO 27001, BIO2 and NIS2.
6. Citizen-relevant information omission in LLM summarization: study which information LLM summaries of long government documents drop, and whether opposition and minority positions are left out more often.

Interested? Send us an expression of interest, or contact us with any questions!