ALP: An authentication and leak prediction model for Cloud Computing privacy

The current paper is an attempt to integrate privacy preservation with privacy leak detection in the context of text mining. This may be considered as extending the Storage as a Service feature of Cloud Computing wherein the user in the role of content creator submits the documents to be stored. There exists a facility to cluster these documents based on the concept-based mining algorithm. The clustered documents are usually available in the form a tree. When a user in the role of a subscriber requests to access a document, this request for access will have to go through an authentication procedure based on the Leakage Free Redactable Signature Scheme. Access control information is being maintained in the form of a Cloud user access control list. A privacy detection leak module which detects privacy leaks depending upon the pattern of previous privacy leaks is also being proposed. This information is then used to update the cloud user access control list and users responsible for privacy leaks are prevented from accessing the cloud service. The current document being requested by the user together with the information from the access control list is used to decide which part of the redacted trees have to be made available to the user as a response. Thus this combination of this authentication procedure and privacy leak detection can be used to ensure the privacy of the sensitive information stored by the user in the cloud.