Researchers at the School of Informatics are exploring ways in which ideas from programming languages can provide formal and general foundations to provide secure, end-to-end provenance support in computer systems.
Provenance issues arise in many areas of online information management. Governmental, military, and business decisions are only as good as the data they are based on, and if this data is noisy, the results may be poor. Provenance is important for enabling decision-makers to distinguish signal from noise, for example to understand when multiple pieces of evidence are duplicates of a single report.
Provenance also has subtle interactions with security, especially where reports of an event may be inaccurate or incomplete, or where actors have incentives to deceive.
Understanding the provenance of data and results of computations is important in a wide variety of contexts, and researchers and developers are considering systematically recording provenance information automatically, for example using the recent W3C standard PROV. However, an ad hoc approach to provenance tracking introduces problems of its own: widespread monitoring may also leak confidential information, violate privacy by capturing user behaviour, yet may not actually go far enough to solve the problem. This project will study the trade-off between ‘capturing enough’ to provide greater confidence in results while not `capturing too much’ and leaking sensitive information or violating users’ privacy, developing formal foundations needed to secure provenance.
The research is part of a four year project funded by EOARD.
For more information see: