![Trust Models based on Subjective Logic
• A trust relationship between A and B is denoted as [A:B]. The transitivity of two
arcs is denoted as “:” and the fusion of two parallel paths is denoted as “◊”.
• The trust network of Figure can be expressed as:
[A,D] = ([A,B]:[B,D]) ◊ ([A,C] : [C,D])
• The corresponding transitivity operator for opinions denoted as and the
corresponding fusion operator as .
• The mathematical expression for combining the opinions about the trust
relationships is:
28](https://image.slidesharecdn.com/cs6010socialnetworkanalysis-unitiv-190410062948/85/CS6010-Social-Network-Analysis-Unit-IV-28-320.jpg)
![Trust Network Analysis - Operators for Deriving Trust
• Subjective logic is a belief calculus specifically developed for modeling trust
relationships.
– Belief functions on binary state spaces are called subjective opinions
• Expressed in the form of an ordered tuple
– where b, d, and u represent belief, disbelief and uncertainty respectively where
b, d, u ∈ [0, 1] and b+d+u = 1
• The base rate parameter a ∈ [0,1] represents the base rate probability in the
absence of evidence, and is used for computing an opinion’s probability
expectation value
31](https://image.slidesharecdn.com/cs6010socialnetworkanalysis-unitiv-190410062948/85/CS6010-Social-Network-Analysis-Unit-IV-31-320.jpg)
![Trust Network Analysis - Trust Path Dependency and
Network Simplification
• A single trust relationship can be expressed as a directed edge between two nodes
that represent the trust source and the trust target of that edge.
• Trust networks can have dependent paths. The expression for the graph on the
right-hand side of Fig. 4.7.a would be: ([A,D]) = (([A,B] : [B,D]) ◊ ([A,C] :
[C,D]))
• The expression for the graph on the left-hand side of Fig. 4.7.a would be: ([A,D])
= (([A,B] : [B,D]) ◊ ([A,C] : [C,D]) ◊ ([A,B] : [B,C] : [C,D]))
• Assuming that the path ([A,B]:[B,C]:[C,D]) is the weakest path in the graph,
network simplification of the dependent graph would be to remove the edge [B,C]
from the graph.
34](https://image.slidesharecdn.com/cs6010socialnetworkanalysis-unitiv-190410062948/85/CS6010-Social-Network-Analysis-Unit-IV-34-320.jpg)
CS6010 Social Network Analysis Unit IV
- 1. CS6010 - Social Network Analysis Unit IV – Predicting Human Behavior & Privacy Issues Kaviya.P AP/IT Kamaraj College of Engineering & Technology 1
- 2. Unit IV – Predicting Human Behavior & Privacy Issues Understanding and predicting human behaviour for social communities - User data management - Inference and Distribution - Enabling new human experiences - Reality mining - Context - Awareness - Privacy in online social networks - Trust in online environment - Trust models based on subjective logic - Trust network analysis - Trust transitivity analysis - Combining trust and reputation - Trust derivation based on trust comparisons - Attack spectrum and countermeasures. 2
- 3. Understanding and predicting human behaviour for social communities 3
- 4. Understanding and predicting human behaviour for social communities • Ubiquitous computing environment will allow everyone permanent access to the Internet anytime, anywhere and anyhow. • QoE (Quality of Experience) is a consequence of a user‟s internal state (e.g., predispositions, expectations, needs, motivation, mood), the characteristics of the designed system (e.g., usability, functionality, relevance) and the context (or the environment) within which the interaction occurs (e.g., social setting, meaningfulness of the activity). 4
- 5. User data management, Inference and Distribution 5
- 6. User data management, Inference and Distribution • For user flexibility and personalization requires user profile management systems which include limited information about user preferences and contexts. • In order to apply user information across a range of services and devices, there is a need for standardization of user related data and the architecture that enables their interoperability. • These efforts have been taken by European Telecommunications Standards Institute (ETSI), the Third Generation Partnership Project (3GPP), Open Mobile Alliance (OMA). 6
- 7. User data management, Inference and Distribution • Common Profile Storage (CPS) is defined by 3GPP. – CPS: A framework for streamlining service-independent user data and storing it under a single logical structure in order to avoid duplications and data inconsistency. • Logically centralized data storage can be mapped to physically distributed configurations and should allow data to be accessed in a standard format. • Data storage can be grouped into three main classes: the syntactic, semantic and modeling approaches which enable interoperability of user profile data management for a Future Internet. 7
- 8. User data management, Inference and Distribution • To improve the degree of services personalization it is important to generate new information from the existing one. • Social networks, user modeling and reality mining techniques can be empowered to study patterns and predict future behaviors. • The basic motivation is the demand to exploit knowledge from various amounts of data collected, pertaining to social behavior of users in online environments. • To handle complex situations, the concept of decomposition is applied to the situation into a hierarchy of sub-situations. • These sub-situations can be handled autonomously with respect to sensing and reasoning. The handling of complex situations can be simplified by decomposition. 8
- 9. User data management, Inference and Distribution • Another perspective is called layered reasoning, – Stage I - feature extraction and grouping (i.e., resulting in low-level context) – Stage II - state and activity recognition (i.e., originating mid-level context) – Stage III - prediction and inference of new knowledge • Research in Social network usually focuses on – Quantifying or qualifying the relationship between peers, – Algorithms such as centrality and prestige can be used to calculate the proximity, influence or importance of a node in a network, – Clustering and classification can be applied to similarity computation, respectively. 9
- 10. User data management, Inference and Distribution • Combining all of pre-enunciated concepts with ontologies and semantic technologies, a generic framework for managing user related data is proposed, will pave the way to understanding and predicting future human behavior within social communities. 10
- 11. Enabling new human experiences - Reality mining, Context-Awareness 11
- 12. Enabling new human experiences • It is important to understand what are the technologies behind it, how to link them and what can they achieve when combined in synergy. Technologies • Social Networks – Social network means ongoing relations among people that matter to those engaged in the group, either for specific reasons or for more general expressions of mutual agreement. – Social networks among individuals who may not be related can be validated and maintained by agreement on objectives, social values, or even by choice of entertainment. – Social networks are trusted because of shared experiences and the perception of shared values and shared needs – Behavior of individuals in online networks can be slightly different from the same individuals interacting in a more traditional social network (reality). 12
- 13. Enabling new human experiences Technologies • Reality Mining – To overcome the differences between online and “offline” networks, reality mining techniques can be empowered to approximate both worlds, proving awareness about people actual behavior. – It is the collection and analysis of machine sensed environmental data pertaining to human social behavior. – It typically analyzes sensor data from mobiles, video cameras, satellites, etc. – Predictive patterns such as “honest signals” provide major factors in human decision making. – Reality mining enables big picture of specific social contexts by aggregating and averaging the collected data. – It allows data/events correlation and consequently future occurrences extrapolation. 13
- 14. Enabling new human experiences Technologies • Context-Awareness – By assessing and analyzing visions and predictions on computing, devices, infrastructures and human interaction, it becomes clear that: • Context is available, meaningful, and carries rich information in such environments. • Users’ expectations and user experience is directly related to context. • Acquiring, representing, providing, and using context becomes a crucial enabling technology for the vision of disappearing computers in everyday environments. 14
- 15. Enabling new human experiences – Architectural Methodology & Framework • To enable human behavior understanding and prediction, there are several independent but complementary steps that can be grouped into three different categories: – Data Management – New Knowledge Generation and – Service Exposure and Control • Data Management – Data acquisition – Gathering information from different information systems. – Data is not usually captured without errors. Therefore it is necessary to preprocess it in advance before mining. – Data is mined by using two different approaches: • know statistical algorithms to help pattern recognition and consequent algorithmic modeling, • the opposite approach, where specific algorithms are designed to identify patterns in the data. 15
- 16. Enabling new human experiences – Architectural Methodology & Framework 16
- 17. Enabling new human experiences – Architectural Methodology & Framework • Knowledge Generation – New information inference is based on user related data (called as context). – Three different categories: Real-time, Historical data, Reasoned context – The main context management features are context acquisition, context aggregation & fusion, context dissemination, discovery and lookup. – In order to manipulate context information, it must be represented in some form that is compatible with the models. – These models could be object oriented, ontological, rule based, logic based, based on semantic graphs or fuzzy logic sets. – Reasoning mechanisms allow high-level context to be deduced or situations to be recognized that is output of one process can be used as an input to another. – It is very important to stress that the prediction does not necessarily anticipates the user wishes or desires, but a possible future that could be interesting for the user. 17
- 18. Enabling new human experiences – Architectural Methodology & Framework • Service Exposure and Control – User-centric and relates to the ability of the user to stay in control of the whole scenario, enabling it to specify when, what, why, who, where and how the data is or can be accessed. – Set of features is associated with the capacity of exposing this information (both raw data and inferred one) to third party service providers (such as advertising agencies), through well defined web service interfaces. – Besides exposing user related information, the human enabler allows data to be subscribed, syndicated or updated on request. 18
- 19. Privacy in Online Social Networks 19
- 20. Privacy in Online Social Networks • Users of the online social networks may share their experiences and opinions within the networks about an item which may be a product or service. • Collaborative filtering system is the most popular method in recommender system. • The task is to predict the utility of items to a particular user based on a database of user rates from a sample or population of other users. • If two people rate a set of items similarly, they share similar tastes. In the recommender system, this information is used to recommend items that one participant likes, to other persons in the same cluster. • Performs poor when there is insufficient previous common rating available between users; known as cold start problem. • To overcome the cold start problem trust based approach to recommendation has emerged. 20
- 21. Privacy in Online Social Networks • Trust could be used as supplementary or replacement of collaborative filtering system. • Trust and reputation systems can be used in order to assist users in predicting and selecting the best quality services. • Binomial Bayesian reputation systems: Ratings expressed in a discrete binary form (i.e) positive (e.g. good) or negative (e.g. bad). • Multinomial Bayesian reputation systems: Provide ratings with discrete graded levels such as e.g. mediocre – bad –average – good – excellent. • Trust models based on subjective logic are directly compatible with Bayesian reputation systems because a bijective mapping exists between their respective trust and reputation representations. • This provides a powerful basis for combining trust and reputation systems for assessing the quality of online services. 21
- 22. Privacy in Online Social Networks • Trust systems can be used to derive local and subjective measures of trust, meaning that different agents can derive different trust in the same entity. • Reputation systems compute scores based on direct input from members in the community which is not based on transitivity • Bayesian reputation systems are directly compatible with trust systems based on subjective logic, they can be seamlessly integrated. This provides a powerful and flexible basis for online trust and reputation management. 22
- 23. Trust in Online Environment 23
- 24. Trust in Online Environment • Trust has become important topic of research in many fields including sociology, psychology, philosophy, economics, business, law and IT. • Trust of two forms: – Reliability trust or Evaluation trust – Decision trust • Evaluation trust can be interpreted as the reliability of something or somebody. It can be defined as the subjective probability by which an individual, A, expects that another individual, B, performs a given action on which its welfare depends. • Decision trust captures broader concept of trust. It can be defined as the extent to which one party is willing to depend on something or somebody in a given situation with a feeling of relative security, even though negative consequences are possible. 24
- 25. Trust Models based on Subjective Logic 25
- 26. Trust Models based on Subjective Logic • Subjective logic is a type of probabilistic logic that explicitly takes uncertainty and belief ownership into account. • Arguments in subjective logic are subjective opinions about states in a state space. • A binomial opinion applies to a single proposition, and can be represented as a Beta distribution. • A multinomial opinion applies to a collection of propositions, and can be represented as a Dirichlet distribution. 26
- 27. Trust Models based on Subjective Logic • Binomial opinions are expressed as where d denotes disbelief in x. When the statement x for example says “David is honest and reliable”, then the opinion can be interpreted as reliability trust in David. • Let us assume that Alice needs to get her car serviced, and that she asks Bob to recommend a good car mechanic. When Bob recommends David, Alice would like to get a second opinion, so she asks Claire for her opinion about David. 27
- 28. Trust Models based on Subjective Logic • A trust relationship between A and B is denoted as [A:B]. The transitivity of two arcs is denoted as “:” and the fusion of two parallel paths is denoted as “◊”. • The trust network of Figure can be expressed as: [A,D] = ([A,B]:[B,D]) ◊ ([A,C] : [C,D]) • The corresponding transitivity operator for opinions denoted as and the corresponding fusion operator as . • The mathematical expression for combining the opinions about the trust relationships is: 28
- 30. Trust Network Analysis • Trust networks consist of transitive trust relationships between people, organizations and software agents connected through a medium for communication and interaction. • Trust network analysis using subjective logic (TNA-SL) takes directed trust edges between pairs as input, and can be used to derive a level of trust between arbitrary parties that are interconnected through the network. • In case of no explicit trust paths between two parties exist; subjective logic allows a level of trust to be derived through the default vacuous opinions. • Complex trust networks must be simplified to series-parallel networks in order for TNA-SL to produce consistent results. 30
- 31. Trust Network Analysis - Operators for Deriving Trust • Subjective logic is a belief calculus specifically developed for modeling trust relationships. – Belief functions on binary state spaces are called subjective opinions • Expressed in the form of an ordered tuple – where b, d, and u represent belief, disbelief and uncertainty respectively where b, d, u ∈ [0, 1] and b+d+u = 1 • The base rate parameter a ∈ [0,1] represents the base rate probability in the absence of evidence, and is used for computing an opinion’s probability expectation value 31
- 32. Trust Network Analysis - Operators for Deriving Trust • A’s opinion about x is denoted as . • Transitivity is used to compute trust along a chain of trust edges. • Assume two agents A and B where A has referral trust in B, denoted by , for the purpose of judging the functional or referral trustworthiness of C. • In addition B has functional or referral trust in C, denoted by . • Agent A can then derive her trust in C by discounting B’s trust in C with A’s trust in B, denoted by . • By using the symbol to designate this operator, we define 32
- 33. Trust Network Analysis - Operators for Deriving Trust • Cumulative Fusion is given by, • The effect of the cumulative fusion operator is to amplify belief and disbelief and reduce uncertainty. 33
- 34. Trust Network Analysis - Trust Path Dependency and Network Simplification • A single trust relationship can be expressed as a directed edge between two nodes that represent the trust source and the trust target of that edge. • Trust networks can have dependent paths. The expression for the graph on the right-hand side of Fig. 4.7.a would be: ([A,D]) = (([A,B] : [B,D]) ◊ ([A,C] : [C,D])) • The expression for the graph on the left-hand side of Fig. 4.7.a would be: ([A,D]) = (([A,B] : [B,D]) ◊ ([A,C] : [C,D]) ◊ ([A,B] : [B,C] : [C,D])) • Assuming that the path ([A,B]:[B,C]:[C,D]) is the weakest path in the graph, network simplification of the dependent graph would be to remove the edge [B,C] from the graph. 34
- 36. Trust Transitivity Analysis • Assume two agents A and B where A trusts B, and B believes that proposition x is true. • Then by transitivity, agent A will also believe that proposition x is true. This assumes that B recommends x to A. • In this approach, trust and belief are formally expressed as opinions. The transitive linking of these two opinions consists of discounting B’s opinion about x by A’s opinion about B, in order to derive A’s opinion about x. – solid arrows - initial direct trust, dotted arrow - derived indirect trust 36
- 37. Trust Transitivity Analysis - Uncertainty Favoring Trust Transitivity • A’s disbelief in the recommending agent B means that A thinks that B ignores the truth value of x. As a result A also ignores the truth value of x. • Uncertainty Favoring Discounting: Let A and B be two agents, where A’s opinion about B’s recommendations is expressed as 𝑊𝐵 𝐴 = 𝑏 𝐵 𝐴 , 𝑑 𝐵 𝐴 , 𝑢 𝐵 𝐴 , 𝑎 𝐵 𝐴 • let x be a proposition where B’s opinion about x is recommended to A with the opinion 𝑊𝑥 𝐵 = 𝑏 𝑥 𝐵, 𝑑 𝑥 𝐵, 𝑢 𝑥 𝐵, 𝑎 𝑥 𝐵 • Let 𝑊𝑥 𝐴:𝐵 = 𝑏 𝑥 𝐴:𝐵 , 𝑑 𝑥 𝐴:𝐵 , 𝑢 𝑥 𝐴:𝐵 , 𝑎 𝑥 𝐴:𝐵 – 𝑏 𝑥 𝐴:𝐵 = 𝑏 𝐵 𝐴 𝑏 𝑥 𝐵 – 𝑑 𝑥 𝐴:𝐵 = 𝑑 𝐵 𝐴 𝑑 𝑥 𝐵 – 𝑢 𝑥 𝐴:𝐵 = 𝑑 𝐵 𝐴 +𝑢 𝐵 𝐴 +𝑏 𝐵 𝐴 𝑢 𝑥 𝐵 – 𝑎 𝑥 𝐴:𝐵 = 𝑎 𝑋 𝐵 • 𝑊𝑥 𝐴:𝐵 - The uncertainty favoring discounting opinion of A. • By using the symbol ⊗ : 𝑊𝑥 𝐴:𝐵 = 𝑊𝐵 𝐴 ⊗ 𝑊𝑥 𝐵 • This operator is associative but not commutative. This means that the combination of opinions can start in either end of the path, and that the order in which opinions are combined is significant. 37
- 38. Trust Transitivity Analysis – Opposite Belief Favoring • A’s disbelief in the recommending agent B. A not only disbelieves in x to the degree that B recommends belief, but she also believes in x to the degree that B recommends disbelief in x, because the combination of two disbeliefs results in belief in this case. • Opposite Belief Favoring Discounting: Let A and B be two agents, where A’s opinion about B’s recommendations is expressed as 𝑊𝐵 𝐴 = 𝑏 𝐵 𝐴 , 𝑑 𝐵 𝐴 , 𝑢 𝐵 𝐴 , 𝑎 𝐵 𝐴 • Let x be a proposition where B’s opinion about x is recommended to A with the opinion 𝑊𝑥 𝐵 = 𝑏 𝑥 𝐵 , 𝑑 𝑥 𝐵 , 𝑢 𝑥 𝐵 , 𝑎 𝑥 𝐵 • Let 𝑊𝑥 𝐴:𝐵 = 𝑏 𝑥 𝐴:𝐵, 𝑑 𝑥 𝐴:𝐵, 𝑢 𝑥 𝐴:𝐵, 𝑎 𝑥 𝐴:𝐵 – 𝑏 𝑥 𝐴:𝐵 = 𝑏 𝐵 𝐴 𝑏 𝑥 𝐵 + 𝑑 𝐵 𝐴 𝑑 𝑥 𝐵 – 𝑑 𝑥 𝐴:𝐵 = 𝑏 𝐵 𝐴 𝑑 𝑥 𝐵 + 𝑏 𝐵 𝐴 𝑑 𝑥 𝐵 – 𝑢 𝑥 𝐴:𝐵 = 𝑢 𝐵 𝐴 + (𝑑 𝐵 𝐴 + 𝑏 𝐵 𝐴 )𝑢 𝑥 𝐵 – 𝑎 𝑥 𝐴:𝐵 = 𝑎 𝑋 𝐵 • 𝑊𝑥 𝐴:𝐵 - opposite belief favoring discounted recommendation from B to A. • By using the symbol ⊗ : 𝑊𝑥 𝐴:𝐵 = 𝑊𝐵 𝐴 ⊗ 𝑊𝑥 𝐵 • This operator models the principle that “your enemy’s enemy is your friend”. 38
- 39. Trust Transitivity Analysis – Base Rate Sensitive Transitivity • Base Rate Sensitive Discounting: The base rate sensitive discounting of a belief 𝑊𝐵 𝐴 = 𝑏 𝐵 𝐴 , 𝑑 𝐵 𝐴 , 𝑢 𝐵 𝐴 , 𝑎 𝐵 𝐴 • by a belief 𝑊𝑥 𝐵 = 𝑏 𝑥 𝐵 , 𝑑 𝑥 𝐵 , 𝑢 𝑥 𝐵 , 𝑎 𝑥 𝐵 • Produces the transitive belief 𝑊𝑥 𝐴:𝐵 = 𝑏 𝑥 𝐴:𝐵 , 𝑑 𝑥 𝐴:𝐵 , 𝑢 𝑥 𝐴:𝐵 , 𝑎 𝑥 𝐴:𝐵 – 𝑏 𝑥 𝐴:𝐵 = 𝐸 𝑤 𝐵 𝐴 𝑏 𝑥 𝐵 – 𝑑 𝑥 𝐴:𝐵 = 𝐸 𝑤 𝐵 𝐴 𝑑 𝑥 𝐵 – 𝑢 𝑥 𝐴:𝐵 = 1 − 𝐸 𝑤 𝐵 𝐴 (𝑏 𝑥 𝐵 + 𝑑 𝑥 𝐵 ) – 𝑎 𝑥 𝐴:𝐵 = 𝑎 𝑋 𝐵 • Probability expectation value E 39
- 40. Combining Trust and Reputation - Trust Derivation based on Trust Comparisons 40
- 41. Combining Trust and Reputation • Reputation systems collect ratings about users or service providers from members in a community. • The reputation centre is then able to compute and publish reputation scores about those users and services. • Multinomial Bayesian systems are based on computing reputation scores by statistical updating of Dirichlet Probability Density Functions (PDF) • A posteriori (i.e. the updated) reputation score is computed by combining a priori (i.e. previous) reputation score with new ratings. 41
- 42. Combining Trust and Reputation 42
- 43. Attack Spectrum and Countermeasures 43
- 44. Attack Spectrum and Countermeasures • The diversity of available OSN platforms opens doors for a variety of attacks on privacy of the users, integrity of their profiles, and the availability of the user- provided contents. 44
- 45. Attack Spectrum and Countermeasures Plain Impersonation • The adversary aims to create fake profiles for real-world users • The success of this attack strongly depends on the authentication mechanisms deployed in the registration process. • Since many OSNs tend to authenticate email addresses by requesting confirmations for the registration emails, this attack can be easily performed if an email address is created in advance. • The consequence of plain impersonation is that the adversary can participate in the OSN applications on behalf of the impersonated user with all damaging consequences for the user. • This attack can be thwarted only through the deployment of stronger authentication techniques 45
- 46. Attack Spectrum and Countermeasures Profile Cloning • The goal of the adversary here is to create a profile for some user that is already in possession of some valid profile in the same network. • From the technical point of view this attack can be realized through the registration of the new profile using the same (or similar) content as the existing one. • This is feasible in most OSN platforms since each profile is associated with some unique administrative id and an email address used during the registration. • A possible solution for OSN providers to prevent profile cloning is to deploy mechanisms that are able to detect similarities between different profiles, in particular with regard to the personal information that is visible to the OSN users. • Since cloned profiles typically have later registration date than the original ones, it should be feasible for the OSN provider to distinguish them and remove from the network. 46
- 47. Attack Spectrum and Countermeasures Profile Hijacking • The goal of the adversary mounting a profile hijacking attack is to obtain control over some existing profile within an OSN platform. • Many OSN platforms protect user access to their own profiles via passwords. • Hence, from the technical point of view profile hijacking is successful if the adversary can obtain passwords of other users. • Users choose weak passwords that can be find via an automated dictionary attack. • OSN providers typically deploy protection against such attacks by restricting the number of login attempts or by using techniques that require human interaction such as CAPTCHAs. 47
- 48. Attack Spectrum and Countermeasures Profile Porting • Profile porting is another type of impersonation where some profile that exists within one OSN platform is cloned into another OSN platform. • Thwarting profile porting is not that easy. • In particular, profile similarity detection tools can still be used but only if they can work across multiple OSN platforms. ID Theft • An adversary mounting the ID theft attack should be able to convince anyone about the ownership of some particular OSN profile. • In this way the adversary can possibly misuse the reputation or expertise of the real profile owner for own benefit, while leaving the owner unaware of the attack. • Thwarting ID theft attacks by technical means seems impossible. • The only solution is to rely on other means of real-world identification such as national identity cards, driver’s licenses, etc. 48
- 49. Attack Spectrum and Countermeasures Profiling • An attack against any target OSN user aiming to collect information about OSN activities or further attributes of that user. • This attack can be typically performed by OSN users, possibly in an automated way, since the collectable information is usually publicly accessible by all OSN users. • The risk of profiling attacks performed by OSN users can be diminished via fine-grained access control and anonymizing techniques. • A solution could be to let users decide whether their activities (e.g. discussion comments) should be kept unlinkable to their profiles. 49
- 50. Attack Spectrum and Countermeasures Secondary Data Collection • Secondary data collection an attack that aims to collect information about the owner of some OSN profile via secondary sources apart of the OSN platform. • A typical example of secondary data collection is to use some Internet search engine to find information that can be linked to the profile owner. • Solution is responsibility of the user to limit information kept in the profile in order to avoid its linkability with secondary sources. Fake Request • An adversary with own OSN profile that sends fake requests to other users. • The dissemination of fake requests can be automated. Since many OSN users tend to accept fake requests, the adversary can simplify access to their profiles and activities and possibly obtain additional information, whose visibility is subject to the available direct or nth-grade connections. • These connections can then be misused for the automated collection and aggregation of information. • Solution: Users behave more responsibly upon accepting new connection requests. 50
- 51. Attack Spectrum and Countermeasures Crawling and Harvesting • The goal of crawling is to collect and aggregate publicly available information across multiple OSN profiles and applications in an automated way. • The adversary is simply interested in collecting as much public information within the OSN platform as possible. • This information can then be misused for different purposes, for example for selling data to marketing agencies, etc. • The adversary simultaneously crawls across different OSN platforms is called harvesting. Typically harvesting results in larger datasets with larger amount on private information about the OSN users. Image Retrieval and Analysis • Image retrieval and analysis, an automated attack aiming to collect multimedia data (images, videos, etc.) available with the OSN platform. • This attack is typically followed by the subsequent analysis via automated pattern recognition tools to find links to the OSN profiles of displayed users. • It may reveal information about friends or colleagues that are not necessarily part of the user’s social network, or information about visited locations (location-tracking) shown on the photographs. 51
- 52. Attack Spectrum and Countermeasures Communication Tracking • Communication tracking, a profiling attack aiming to reveal information about communications of the same user. • The attacker may collect more information about the user than available in the profile. • This attack can be mounted in an automated way by searching for comments left by the target user in various OSN applications. Fake Profiles and Sybil Attacks • OSN platforms lack of proper authentication such creation of fake profiles becomes easy. • On the technical side, the user has only to create a new email for the registration of a fake account. • Fake profiles pave the way for Sybil attacks that may serve different purposes. • Owners of fake profiles can establish new connections without disclosing their real identities. This includes distribution of spam messages or other illicit content such as malware and phishing links, illegal advertisement, etc. • Solution for OSN providers to recognize fake profiles is to use IP traceback. • Stronger identification and authentication mechanisms for admission of new users would offer a better protection. 52
- 53. Attack Spectrum and Countermeasures Group Metamorphosis • Group metamorphosis, an attack where group administrators change the group subject to persuade own interests, e.g. political. • Other OSN users who joined the group earlier may remain unaware of this change, which in turn may have negative impact on their reputation. • A possible solution for OSN providers to thwart group metamorphosis attacks is to restrict control of administrators over the interest groups, in particular to prevent them from modifying any information that may have impact on the group as a whole. Ballot Stuffing and Defamation • Ballot stuffing, an attacker wishes to increase public interest to some target OSN user. • This attack may increase the amount of personal messages or connection requests received by the target user resulting in a DoS attack on the physical resources of the OSN user. • On the other hand, ballot stuffing may increase popularity of the profile belonging to the attacker. This can be achieved through recommendations submitted by the attacker using fake profiles. • Defamation attacks aim at decreasing public interest of a target user, in particular by tarnishing the reputation of the Latter • In particular, defamation may lead to blacklisting of the user in contact lists of other users and keep the user away from participation in communication applications such as shared interest groups and discussion forums. 53
- 54. Attack Spectrum and Countermeasures Censorship • OSN providers typically have control over the whole data available within the network. They can deliberately manipulate the user-provided information and contents. • Censorship when applied without substantial reasons may have negative impact on the OSN users. • For example, in OSN platforms focusing on business contacts users often advertise their expertise. In this scenario censorship may be misused to favor some users over their competitors. It can be performed by active modification of user-provided contents, which might remain unnoticed by the user. Higher impact can be achieved through the target manipulation of search engines within the network. Collusion Attack • Several users join their malicious activities in order to damage other OSN users or mount attacks against applications of the OSN platform. • In particular, colluding users may start defamation or ballot stuffing campaigns, increase each over reputations, bias the outcome of public polls or influence public discussions. 54