Profil utilisateur | EPFL Graph Search

Un profil utilisateur ou modèle d'utilisateur est un ensemble de données et métadonnées fonction d'un ou plusieurs utilisateurs qui influence le comportement d'un dispositif informatique. Un profil peut être relatif à une personne seule, ou à un groupe de personnes ayant des points communs, tels que par exemple les membres d'un groupe de travail. Utilisations Certaines des données des profils utilisateurs sont données par les utilisateurs eux-mêmes pour la configuration des logiciels et applications qu'ils utilisent. D'autres proviennent de l'utilisation collective de systèmes informatiques connectés en réseau. Le profil utilisateur peut notamment contenir les préférences linguistiques. Mais le profil utilisateur peut être beaucoup plus large, et contenir des préférences booléennes pouvant conduire à un traitement différencié et être issu de calculs probabilistes exploitant par exemple les réseaux bayésiens Les profils utilisateurs sont utilisés dans les logiciels colle

The Role of Compromised Accounts in Social Media Manipulation

Tugrulcan Elmas

In recent years we have seen a marked increase in disinformation including as part of a strategy of so-called hybrid warfare. Adversaries not only directly spread misleading content but manipulate social media by employing sophisticated techniques that exploit platform vulnerabilities and avoid detection. It is getting increasingly important to analyze social media manipulation to better understand, detect and defend public dialogue against it.In this thesis, we contribute to the research on social media manipulation by describing and analyzing how adversaries employ compromised social media accounts. We begin by providing a background of social media: we describe the mechanisms and the influence of the platforms to better understand why the adversaries target them. We then give a detailed overview of social media manipulation, and the techniques to detect and counter it. Next, we discuss our contributions in this thesis: 1) an extensive analysis of an attack on social media algorithms using compromised accounts, 2) a study of the implications of compromised bots for bot research through the characterization of retweet bots, 3) a detection method to find compromised accounts that are later repurposed.Firstly, we uncover and analyze a previously unknown, ongoing astroturfing attack on the popularity mechanisms of social media platforms: ephemeral astroturfing attacks. In this attack, a chosen keyword or topic is artificially promoted by coordinated and inauthentic activity to appear popular. Crucially, this activity is removed as part of the attack which facilitates using compromised accounts that are still managed by their original owners. We detected over 19,000 unique fake trends promoted by over 108,000 accounts. Trends astroturfed by these attacks account for at least 20% of the top 10 global trends. We created a Twitter bot to detect the attacks in real-time and inform the public. Secondly, we study the implications of compromised accounts to bot research. We do this by characterizing retweet bots that have been uncovered by purchasing retweets from the black market. We determine that those accounts were compromised as they observe anomalous behavior, or self-state that they are hacked. We then analyze their differences from human-controlled accounts. From our findings on the nature and life-cycle of retweet bots, we point out several inconsistencies between the retweet bots used in this work and bots studied in prior works. Our findings challenge some of the fundamental assumptions related to bots and in particular how to detect them.Thirdly, we define, describe, and provide a detection method for misleading repurposing, in which an adversary changes the identity of a potentially compromised social media accounts via, among other things, changes to the profile attributes in order to use them for a new purpose while retaining their followers. We propose a methodology to flag repurposed accounts, and detected over 100,000 such accounts. We also characterize repurposed accounts and found that they are more likely to be repurposed after a period of inactivity and deleting old tweets. We present a tool to root out accounts that became popular and repurposed later.Our work is significant in presenting how breaches of user security jeopardize platform security and public dialogue. Furthermore, it enhances the knowledge of how the bots and troll accounts work and aid platforms and researchers in building new solutions.

EPFL2022

TEE-based decentralized recommender systems: The raw data sharing redemption

Akash Balasaheb Dhasade, Nevena Dresevic, Anne-Marie Kermarrec, Rafael Pereira Pires

Recommenders are central in many applications today. The most effective recommendation schemes, such as those based on collaborative filtering (CF), exploit similarities between user profiles to make recommendations, but potentially expose private data. Federated learning and decentralized learning systems address this by letting the data stay on user's machines to preserve privacy: each user performs the training on local data and only the model parameters are shared. However, sharing the model parameters across the network may still yield privacy breaches. In this paper, we present REX, the first enclave-based decentralized CF recommender. REX exploits Trusted execution environments (TEE), such as Intel software guard extensions (SGX), that provide shielded environments within the processor to improve convergence while preserving privacy. Firstly, REX enables raw data sharing, which ultimately speeds up convergence and reduces the network load. Secondly, REX fully preserves privacy. We analyze the impact of raw data sharing in both deep neural network (DNN) and matrix factorization (MF) recommenders and showcase the benefits of trusted environments in a full-fledged implementation of REX. Our experimental results demonstrate that through raw data sharing, REX significantly decreases the training time by 18.3x and the network load by 2 orders of magnitude over standard decentralized approaches that share only parameters, while fully protecting privacy by leveraging trustworthy hardware enclaves with very little overhead.

2022

Contextualized and personalized location-based services

Shijun Yu

Advances in the technologies of smart mobile devices and tiny sensors together with the increase in the number of web resources open up a plethora of new mobile information services where people can acquire and disseminate information at any place and any time. Location-based services (LBS) are characterized by providing users with useful and local information, i.e. information that belongs to a particular domain of interest to the user and can be of use while the user remains in a particular area. In addition, LBS need to take into account the interactions and dependencies between services, user and context for the information filtering and delivery in order to fulfill the needs and constraints of mobile users. We argue that consequently it brings up a series of technical challenges in terms of data semantics and infrastructure, context-awareness and personalization, as well as query formulation and answering etc. They can not be simply extended from existing traditional data management strategies. Instead, they need a new solution. Firstly, we propose a semantic LBS infrastructure on the basis of the modularized ontologies approach. We elaborate a core ontology which is mainly composed of three modules describing the services, users and contexts. The core ontology aims at presenting an abstract view (a model) of all information in LBS. In contrast, data describing the instances (of services user and actual contextual data) are stored in three independent data stores, called the service profiles, user profiles and context profiles. These data are semantically aligned with the concepts in the core ontology through a set of mappings. This approach enables the distributed data sources to be maintained in a autonomous manner, which is well adapted to the high dynamics and mobility of the data sources. Secondly, we separately address the function, features, and our modelling approach of the three major players, i.e. service, context and user in LBS. Then, we define a set of constructs to represent their interactions and inter-dependencies and illustrate how these semantic constructs can contribute to personalized and contextualized query processing. Service classes are organized in a taxonomy, which distinguishes the services by their business functions. This concept hierarchy helps to analyze and reformulate the users' queries. We introduce three new kinds of relationships in the service module to enhance the semantics of interactions and dependencies between services. We identify five key components of contexts in LBS and regard them as a semantic contextual basis for LBS. Component contexts are related together by specific composition relationships that can describe spatio-temporal constraints. A user profile contains personal information about a given user and possibly a set of self-defined rules, which offer hints on what the user likes or dislikes, and what could attract him or her. In the core ontology clustering users with common features can help the cooperative query answering. Each of the three modules of the core ontology is an ontology in itself. They are inter-related by relationships that link concepts belonging to two different modules. The LBS fully benefits from the modularized structure of the core ontology. It allows restricting the search space, as well as facilitating the maintenance of each module. Finally, we studied the query reformulation and processing issues in LBS. How to make the query interface tangible and provide rapid and relevant answers are typical concerns in all information services. Our query format not only fully obeys the "simple, tangible and effective" golden-rules of user-interface design, but also satisfies the needs of domain-independent interface and emphasizes the importance of spatio-temporal constraints in LBS. With pre-defined spatio-temporal operators, users can easily specify in their queries the spatio-temporal availability they need for the services they are looking for. This allows eliminating most of irrelevant answers that are usually generated by keyword-based approaches. Constraints in the various dimensions (what, when, where and what-else) can be expressed by a conjunctive query, and then be smoothly translated to RDF-patterns. We illustrate our query answering strategy by using the SPARQL syntax, and explain how the relaxation can be done with rules specified in the query relaxation profile.

EPFL2008