This is Penn State

Secrecy-Preserving Inference and Query Answering

(Funded in part by a grant from the National Science Foundation)
 

Productive interaction and collaboration among business partners, different governmental agencies (e.g., intelligence, law enforcement, public policy), or independent nations acting on matters of global concern (e.g., counter-terrorism, international finance) requires the need to share information to be balanced against the need to protect sensitive or confidential information from unintended disclosure. We aim to address this need by developing the theoretical foundations of, and algorithms and software for secrecy-preserving reasoning, that is, the process of answering queries against knowledge bases that include secret knowledge, based on inference that may use secret knowledge without revealing it. The privacy-preserving reasoning framework introduced by us in Bao et al. (2007) offered one of the first semantics-based approaches to secrecy-preserving query answering in the simple case of hierarchical knowledge bases. Work in progress is focused on:

  • The development of theoretical foundations of secrecy-preserving reasoning leading to general strategies for transforming sound and complete reasoners for knowledge bases into provably secrecy-preserving reasoners that differentially tradeoff the informativeness of the reasoning against the computational overhead of secrecy preservation (relative to reasoners that are oblivious to secrecy).
  • The development of secrecy-preserving reasoners for a broad class of knowledge bases of practical interest in networked information systems including hierarchical, propositional, RDF (resource description framework) and description logic (DL) knowledge bases.
  • Extensions of the secrecy-preserving reasoning algorithms to settings with multiple querying agents, under various restrictions on communication between the agents.
  • Modular, open source implementations and experimental evaluation of secrecy-preserving reasoners.