Reliability score inference and recommendation using fuzzy-based technique for social media applications

Abstract

Usage of online social networks (OSN) has been gaining momentum in the recent past. People with varying cultural background and age, spanning the globe, share their views and day-to-day affairs with their community. There are an abundant number of internal applications as well as third-party applications (TPA) available to the users either through these OSN sites or that which uses the OSNs identity service to login and access the service. When users wish to use the TPA, the user’s data has to be shared with the TPA’s server. Otherwise, the service would be denied. When users share such information with the TPA, firstly, the users may not be aware of the reliability level of TPA and how those TPA would be handling the user’s data. Secondly, they need not provide the promised level of service but still would have acquired the data from the users. Hence, it is necessary to check the TPA’s level of service and the data requested by them before using the service. The reliability inference application (RIA) and application recommender proposed in this work are based on fuzzy inference mechanism. They mimic the human expert’s decision of choosing a TPA. The RIA trades off the risk associated with sharing the data with the level of service offered and renders the reliability score of the applications. The application recommender presents the users with the recommendation for TPA as highly recommended, recommended or recommended with risk based on the user’s privacy preference. It assists the user to choose a TPA that provides the desired level of service matching the user’s privacy preferences.