Detection of privacy leaks in social networks: a vision of the estate of art

Abstract

When talking about social networks, all the advantages they offer to the user in terms of communication and interaction with the world are expressed, and it is also recognized that they are a source of extraction and use of data for various purposes, although some with illicit purposes. Social networks have become popular and people dedicate daily time to review the main notifications received, as well as communicate, publish, comment or place labels. The need to control privacy is fundamental, but the configurations that can be made to avoid it are not enough. The user posts messages that contain elements of natural language, but that are difficult to interpret by software. To perform a search for items considered private that are found in the published texts or that can be inferred by the total revision of the visible information published in the user's profile, it is necessary to process the text to clean it, normalize it and classify it. Private information can easily be inferred, just by associating the different textual publications that were shared by the user or another person. It is opportune to make use of text mining techniques to avoid the publication of private information in the profile that is shared with the group of friends of the social network.