Information Retrieval as a Service for Multiple Heterogeneous Data-Privacy Model

Abstract

The problem of information retrieval service from multiple heterogeneous autonomous and independent data sources is a rapidly growing requirement because of the proliferation of the internet and the large data thereon. With the emergence of cloud computing along with the great amount of database research on data integration and querying from multiple heterogeneous data sources through mediation [1], much work done on security in databases and privacy preserving query processing [2]. Also the development of the concept of cloud database [3] the platform is perfectly set for information retrieval as a service (IRaaS) as another highly prospective cloud based service. Such services in the current scenario must meet the basic characteristics of cloud computing such as scalability, location independent resource pooling on demand and ubiquitous network access and thus should be hosted in a cloud. IRaaS is a cloud based service in general and SaaS in particular. According to objective and orientation the customers of IRaaS could vary a lot. The services given to the customers who are interested in arbitrary queries are basically referred as open access IR services, where the service provider has to retrieve data from dynamically selected data sources and needs to be extremely powerful, something like a combination of the Google search engine, YouTube and some other database search engines. On the other hand there are more realistic closed access IR services which are applications or business specific such as services meant for commercial banks, tour and travel, crime record etc. We propose different service models such as open access, closed access, collaborative IRaaS and enterprise IRaaS, and have provided the taxonomy of IRaaS.

For an IR service to be successful and attractive to the information seekers as well as to the information providers, the privacy maintenance is a core issue. We have modelled the overall privacy requirement of an IR service as a combination of four types of privacy protection, namely, identity protection, data protection, query protection and result protection. We have proposed a detailed privacy model and a secure framework sketch for the closed access service which can be used for a privacy preserving information retrieval system.