Information Retrieval as a Service – IRaaS : A Concept Paper on Privacy Analysis

Abstract

Privacy analysis has not always got proper attention in the literature often overtaken by security algorithms. This work attempts to fill in this gap. The strength of this work we believe is in the privacy analysis conducted in depth for a complex problem following an objective method. Providing information retrieval service from multiple heterogeneous autonomous data sources is a natural requirement. Queries may appear within an application framework or may be ad hoc. The service must meet the basic characteristics of cloud computing such as scalability and ubiquitous network access. We recognize this as IR as a Service (IRaaS), and propose different models - Open Access, Closed Access, Collaborative and Enterprise IRaaS. Such services should win the trust of all participants by enforcing required security as per privacy requirements. Explicit privacy statement going into depth on the need of sharing of information and computation which is agreed by each party helps secure computation through developing mutual trusts for the entire process of computation and communication. Privacy need of IRaaS is modelled by six privacy issues: identity, schema, data, result, query and query distribution, each issue analyzed on privacy protections involving information sharing among the parties. To limit flexibility we apply dominance relations between privacy issues across privacy protections and vice versa. We suggest a privacy algebra and apply it to IRaaS for consolidation. Finally a skeleton IR framework for IRaaS with an algorithm for secure query processing using the privacy model is proposed.