With an abundance of social network data being released, the need to protect sensitive information within these networks has become an important concern of data publishers. In this thesis we focus on the popular notion of k-anonymization as applied to social network graphs. Given such a network N, the problem we study is to transform N to N', such that some property P of each node in N' is attained by at least k-1 other nodes in N'. We study edge-labeled, vertex-labeled and unlabeled graphs, since instances of each occur in real-world social networks. Our main contributions are as follows 1. When looking at edge additions, we show that k-label sequence anonymity of arbitrary edge-labeled graphs is NP-complete, and use this fact to prove hardness results for many other recently introduced notions of anonymity. We also present interesting hardness results and algorithms for labeled and unlabeled bipartite graphs. 2. When looking at node additions, we show that on vertex-labeled graphs, the problem is NP-complete. For unlabeled graphs, we give an efficient (near-linear) algorithm and show that it gives solutions that are optimal modulo k, a guarantee that is novel in the literature. We examine anonymization both from its theoretical foundations and empirically, showing that our proposed algorithms for anonymization maintain structural properties shown to be necessary for graph analysis. / Graduate

Consumer electronics such as cellular phones and portable computers with short-range communication capabilities have enabled the large-scale information dissemination through user mobility and contact, without the assistance of communication infrastructures. In such a new communication paradigm, one challenge is to determine when and how to forward a message to the destination, possibly through a series of third parties. This problem has attracted a lot of attention in the literature lately, with proposals ranging from epidemic to single or multi-copy spray and wait or focus strategies. However most existing work assumed independent or identically distributed mobility. Observing most human mobility and interaction are interest-driven in the real world, in this research, we evaluate the performance of these schemes with an interest-driven mobility model. We further propose to take the user interest into account when determining routing strategies to further improve the performance of these schemes for mobile social networks. Simulation results have demonstrated the efficacy of the interest-aware routing strategies. / Graduate

Improving road safety and traffic efficiency has been a long-term endeavor for the government, automobile industry and academia. Recently, the U.S. Federal Communication Commission (FCC) has allocated a 75 MHz spectrum at 5.9 GHz for vehicular communications, opening a new door to combat the road fatalities by letting vehicles communicate to each other on the roads. Those communicating vehicles form a huge Ad Hoc Network, namely Vehicular Ad Hoc Network (VANET). In VANETs, a variety of applications ranging from the safety related (e.g. emergence report, collision warning) to the non-safety related (e.g., delay tolerant network, infortainment sharing) are enabled by vehicle-to-vehicle (V-2-V) and vehicle-to-roadside (V-2-I) communications. However, the flourish of VANETs still hinges on fully understanding and managing the challenging issues over which the public show concern, particularly, security and privacy preservation issues. If the traffic related messages are not authenticated and integrity-protected in VANETs, a single bogus and/or malicious message can potentially incur a terrible traffic accident. In addition, considering VANET is usually implemented in civilian scenarios where locations of vehicles are closely related to drivers, VANET cannot be widely accepted by the public if VANET discloses the privacy information of the drivers, i.e., identity privacy and location privacy. Therefore, security and privacy preservation must be well addressed prior to its wide acceptance. Over the past years, much research has been done on considering VANET's unique characteristics and addressed some security and privacy issues in VANETs; however, little of it has taken the social characteristics of VANET into consideration. In VANETs, vehicles are usually driven in a city environment, and thus we can envision that the mobility of vehicles directly reflects drivers' social preferences and daily tasks, for example, the places where they usually go for shopping or work. Due to these human factors in VANETs, not only the safety related applications but also the non-safety related applications will have some social characteristics. In this thesis, we emphasize VANET's social characteristics and introduce the concept of vehicular social network (VSN), where both the safety and non-safety related applications in VANETs are influenced by human factors including human mobility, human self-interest status, and human preferences. In particular, we carry on research on vehicular delay tolerant networks and infotainment sharing --- two important non-safety related applications of VSN, and address the challenging security and privacy issues related to them. The main contributions are, i) taking the human mobility into consideration, we first propose a novel social based privacy-preserving packet forwarding protocol, called SPRING, for vehicular delay tolerant network, which is characterized by deploying roadside units (RSUs) at high social intersections to assist in packet forwarding. With the help of high-social RSUs, the probability of packet drop is dramatically reduced and as a result high reliability of packet forwarding in vehicular delay tolerant network can be achieved. In addition, the SPRING protocol also achieves conditional privacy preservation and resist most attacks facing vehicular delay tolerant network, such as packet analysis attack, packet tracing attack, and black (grey) hole attacks. Furthermore, based on the ``Sacrificing the Plum Tree for the Peach Tree" --- one of the Thirty-Six Strategies of Ancient China, we also propose a socialspot-based packet forwarding (SPF) protocol for protecting receiver-location privacy, and present an effective pseudonyms changing at social spots strategy, called PCS, to facilitate vehicles to achieve high-level location privacy in vehicular social network; ii) to protect the human factor --- interest preference privacy in vehicular social networks, we propose an efficient privacy-preserving protocol, called FLIP, for vehicles to find like-mined ones on the road, which allows two vehicles sharing the common interest to identify each other and establish a shared session key, and at the same time, protects their interest privacy (IP) from other vehicles who do not share the same interest on the road. To generalize the FLIP protocol, we also propose a lightweight privacy-preserving scalar product computation (PPSPC) protocol, which, compared with the previously reported PPSPC protocols, is more efficient in terms of computation and communication overheads; and iii) to deal with the human factor -- self-interest issue in vehicular delay tolerant network, we propose a practical incentive protocol, called Pi, to stimulate self-interest vehicles to cooperate in forwarding bundle packets. Through the adoption of the proper incentive policies, the proposed Pi protocol can not only improve the whole vehicle delay tolerant network's performance in terms of high delivery ratio and low average delay, but also achieve the fairness among vehicles. The research results of the thesis should be useful to the implementation of secure and privacy-preserving vehicular social networks.

The resources and localization abilities available in modern smartphones have provided a huge boost to the popularity of location-based applications. In these applications, users send their current locations to a central service provider and can receive content or an enhanced experience predicated on their provided location. Privacy issues with location- based applications can arise from a central entity being able to store large amounts of information about users (e.g., contact information, attributes) and locations (e.g., available businesses, users present). We propose an architecture for a privacy-friendly location hub to encourage the development of mobile location-based social applications with privacy- preserving features. Our primary goal is to store information such that no entity in our architecture can link a user’s identity to her location. We also aim to decouple storing data from manipulating data for social networking purposes. Other goals include designing an architecture flexible enough to support a wide range of use cases and avoiding considerable client-side computation. Our architecture consists of separate server components for storing information about users and storing information about locations, as well as client devices and optional com- ponents in the cloud for supporting applications. We describe the design of API functions exposed by the server components and demonstrate how they can be used to build some sample mobile location-based social applications. A proof-of-concept implementation is provided with in-depth descriptions of how each function was realized, as well as experi- ments examining the practicality of our architecture. Finally, we present two real-world applications developed on the Android platform to demonstrate how these applications work from a user’s perspective.

国立情報学研究所で電子化したコンテンツを使用している。

The study of preschool children's peer relations has primarily employed sociometric interviews and observations of children's social participation during free-play situations as methods of assessment. These assessments of peer relations have traditionally demonstrated lack of fit between attitude (as measured by sociometry) and behavior (observation of social participation). Often, sociometric peer preference and acceptance does not correspond to children's observation of social participation during freeplay. The present study was designed to improve this attitude-behavior fit by developing the Crystallized Sociometric Scales which employed post stratification of opinion responses. Subjects consisted of 65 preschool children between three- to five- years- of age, divided into two preschool groups of 45 and 20 subjects. Interviews on both the traditional sociometric scales and the crystallized sociometric (weighted with questions on peer exclusivity and friendship concept) were obtained. Observation of social participation used Parten's category of play with interval time sampling. Sociometric analyses of data provided a child by child picture of the social structure of the peer group. The discrepancy scores between social participation and the traditional and crystallized sociometric variables were compared using t-tests. Results indicated that peer exclusivity provided the best attitude-behavior fit and was significantly better than the traditional nomination score. The level of the friendship concept did not affect the social participation of children. Crystallized ratings were not significantly different from the traditional ratings. The stability of the Crystallized Sociometric Scales was moderate, although it was significantly better than the traditional sociometric scales. It was concluded that crystallized sociometric nominations provide a methodology to improve the attitude-behavior fit. Although correlated, sociometry and social participation measure different aspects of peer relations whose relationship can be influenced by measurement procedures. / Graduation date: 1992

The resources and localization abilities available in modern smartphones have provided a huge boost to the popularity of location-based applications. In these applications, users send their current locations to a central service provider and can receive content or an enhanced experience predicated on their provided location. Privacy issues with location- based applications can arise from a central entity being able to store large amounts of information about users (e.g., contact information, attributes) and locations (e.g., available businesses, users present). We propose an architecture for a privacy-friendly location hub to encourage the development of mobile location-based social applications with privacy- preserving features. Our primary goal is to store information such that no entity in our architecture can link a user’s identity to her location. We also aim to decouple storing data from manipulating data for social networking purposes. Other goals include designing an architecture flexible enough to support a wide range of use cases and avoiding considerable client-side computation. Our architecture consists of separate server components for storing information about users and storing information about locations, as well as client devices and optional com- ponents in the cloud for supporting applications. We describe the design of API functions exposed by the server components and demonstrate how they can be used to build some sample mobile location-based social applications. A proof-of-concept implementation is provided with in-depth descriptions of how each function was realized, as well as experi- ments examining the practicality of our architecture. Finally, we present two real-world applications developed on the Android platform to demonstrate how these applications work from a user’s perspective.

This thesis explores several issues in the adaptation process of immigrants and their children in Canada. Chapter 2 investigates why second-generation immigrants are better educated than the remaining population. Using a standard human capital framework where individuals choose how much to invest in both their children's and their own human capital, I show that a gap in education can arise in the absence of differences in unobservable characteristics between immigrants and the native born. Rather, it can arise due to institutional factors such as imperfect transferability of foreign human capital and credit constraints. The model's key implication is a negative relationship between parental human capital investments and children's educational attainment, particularly in families with uneducated parents. I find strong empirical evidence of such tradeoffs in human capital investments occurring within immigrant families. Chapter 3 re-assesses the effect of living in an ethnic enclave on labour market outcomes of immigrants. I find evidence of cohort effects in the relationship between mean earnings and the proportion of co-ethnics in the CMA which vary by education level. Next, using information on the proportion of one's friends who share one's ethnicity, I test a common assumption that the enclave effect is a network effect. I find that traditional, geography-based measures of the ethnic enclave effect capture the impact of factor(s) other than social networks. In fact, the two effects generally offset each other to some degree in determining immigrant employment outcomes. Neither measure has a statistically significant effect on average immigrant earnings, at least in cross-sectional data. Chapter 4, co-authored with David Green and Craig Riddell, tests two alternative theories about why immigrants earn less than native-born workers with similar educational attainment and experience - discrimination versus lower skills (measured by literacy test scores). We find that immigrant workers educated abroad have lower cognitive skill levels (assessed in English or French) than similar native-born workers. This skills gap can explain much of the earnings gap. At the same time, foreign-educated immigrants receive no lower returns to skills than the native born. These results offer strong evidence against the discrimination hypothesis.

Thesis (Ph.D.)--Ohio University, August, 2007. / Title from PDF t.p. Includes bibliographical references.

Thesis (MPH.--Faculty of Health Sciences)-University of Pretoria, 2007. / Includes bibliographical references. Also available online.