This project proposes the design and the sperimentation of the MISSION system (My Information Shield for Social Networks). Such a system intends to automatically identify the information received by users of a social network to help them to defend themselves against threats like scam, cyberbullying or sexual predation, and to avoid access to suspicious, misleading, and fake data.
Social Network Sites (SNSs) are an ideal place for Internet users to keep in touch, share information about their daily activities and interests, publishing and accessing documents, photos and videos. SNSs like Facebook, Twitter or Google+ give the ability to create a semi-public profile, to have a list of peers to interact with and to post and to read what others have posted.
According to Alexa's statistics, Google and Facebook are the first websites for exchanged traffic, while Twitter is anyway within the top ten: together they count around 2 billion of users. Unfortunately, SNSs are also the ideal plaza for proliferation of harmful information. Scam, cyberbullying, sexual predation, self-harm practices, deceptive reviews, fake accounts and links to malware are the effective results of the dissemination of malicious information on SNSs. Although the few experienced users could be able to face threats and distinguish between deceptive and honest content, such a challenge cannot be easily won by the great majority of users: Media frequently report evidences about the consequences that naïve and quite emotive users have faced to.
Our commitment is, therefore, to limit the harmfulness of a certain kind of information exchanged on SNSs, providing an effective tool that can help all social network users. MISSION provides a software system able to identify the harmful and threatening contents users are exposed to, that will be clearly tagged to help the user to recognize the threat she is involved in.
Until today, no work has combined together user connections, message exchange, and the social profile characteristics to recognize a set of threats, in addition to the cyberbullying, like, e.g., scam, deceptive opinions, sexual predation trials [1] and self-harm practice incitements [2].
The framework resulting from the MISSION project will be characterized by several self-consistent components that can be potentially employed in different contexts. For example, the profile linker could be used inside schools to link bullying accounts to real students, in order to stop potentially harmful behaviors. Similarly, the corpus of characterized threats used to train the anomaly detection system could be used as a dataset for further experiments, to improve the threat detection ability. Overall, since the framework components come from many field of Computer Science, they will pave the way for technology advancements in several directions from machine-learning to anomaly detection systems. The possibility to help the user to report abuses or inappropriate content directly from the MISSION application/plugin to the authority is a feature that makes easier for the users to protect against threats on SNSs. Moreover, this feature facilitates the adoption of the Italian self-regulatory code [3], signed by the Italian Ministry of Economic Development and several SNS's operators (including, Microsoft and Google).
One of the main innovations will be a new modeling framework in which the profiles of SNSs' users are the entity that interact among them and manage the data: there will be a formal definition between the possible operations and the interactions that the users can perform in their environment. The MISSION framework will make use of primitives and rules to allow manipulating and analyzing operation and interaction sequences, in order to permit formal definition and reasoning about equivalences between different interaction sequences and between the different SNSs. Once the behavior and interaction model will be defined, MISSION will have a generic tool to describe any possible type of action that the users can perform in the SNSs, like access to a page, write or receive a message, and so on.
The successful outcome of MISSION will yield a main social impact: users will increase their confidence in the SNSs and in the type of online information they should take care of. The user's cognitive growth occurs with the usage of the framework, which, protecting her from the threats, also teaches her how to ask for help. Then, the users will gradually better understand social networks' relationships and the risks of the information exchanged, becoming aware of the wrong use of the SNSs, of their power to better select their acquaintances and to report to whom concerns all the recognized cases of abuse.
[1] Kontostathis, A.: Chatcoder: Toward the tracking and categorization of Internet predators. In: Text Mining Workshop of Siam Data Mining (SDM 2009)
[2] Chattopadhyay, S. et al. Suicidal Risk Evaluation Using a Similarity-Based Classifier. (2008) In LNCS Advanced Data Mining and Applications
[3] CODICE DI AUTOREGOLAMENTAZIONE PER LA PREVENZIONE E IL CONTRASTO DEL CYBERBULLISMO, http://www.sviluppoeconomico.gov.it/images/stories/documenti/codice_cybe...