This study looks at the importance of social networks amongst urban migrants in Johannesburg. The aim of the study was to look at how the social networks of international migrants function, and how migrants make use of these networks in an urban setting; examining whether this differs between migrants with established social networks available at their final destination before departure, and migrants without such social networks, and if so, how it differs. The study looks at the importance of social networks throughout the entire migration process, more specifically investigating their impact on the decision-making, journey, arrival and adaptation conditions. The literature review highlights research within the areas of social capital and social networks, research which was used to develop the definition of social networks used in this thesis. The literature review further shows that most research on the subject has emphasised the advantages of having friends or family available at the country of destination. This chapter also establishes a set of important indicators which formed the framework of areas which needed to be included in the analysis of adaptation. In order to analyse the importance of social networks, a mixed methods approach was adopted. This approach allowed the quantitative section to establish particular relationships between variables, while the qualitative section explained these relationships further. The comprehensive quantitative data which was used came from the African Cities Project (ACP) which was a comparative and longitudinal survey conducted in 2008. To further explore the findings from this data, a case study was conducted using in-depth interviews with the most interesting migrant group identified in the ACP data; the Somalis. The decision to select the Somalis as the subject of the qualitative case study was based on the findings of the quantitative analysis, and in particular the fact that the Somali respondents in many ways contradicted much of the previous literature on social networks. The findings of this thesis suggest that the significance of social networks during the migration process has often been exaggerated in the literature. According to the data used in this study, migrants without social networks tend to be more successful in many areas, especially when it comes to adapting to the new country. Having personal networks at the country of destination before departure seems to be less important than the cultural knowledge needed to find and make use of the networks and assistance available.

ABSTRACT This research investigates how language affects the livelihoods of migrants and their use of social networks for income generation in Johannesburg. The study focuses on the Cameroonians, a community constituted by two main linguistic groups: Francophones and Anglophones. It explores the historic process of migration into the Republic of South Africa, the motivations behind the choice of the inner city of Johannesburg by Cameroonians, and it differentiates the type of social networks in which Cameroonians in this study engage in. The research report also explores how language shapes social networks among Cameroonians living in the inner city of Johannesburg. Findings of this study reveal that language does not have a direct effect on livelihoods as almost all Cameroonians migrants interviewed in this study are able to speak both English and French. It nevertheless, shapes the respondents’ opportunities and strategies as many rely on linguistic communities to access information, get social assistance, and find jobs. This study found that Cameroonians living in the inner city of Johannesburg use intra-linguistic social networks more than inter-linguistic group social networks for their income generating strategies. It also demonstrates the importance that language has in the lives of these respondents. The study is based on non-representative sample of Cameroonians living in Johannesburg. It used a multi-methods approach including in-depth face-to-face interviews and written sources such as journals, books and research reports were combined to gather relevant data. Thematic content analysis was used to analyse the data.

Thesis advisor: Candace Jones / Management scholars have rarely analyzed how prior social networks might help or hinder the job performance of new organizational members. However, internal and external job markets are increasingly characterized by high mobility of experienced professionals, who have extensive social networks rooted in their past collaborations and shared work experiences. Organizations rely more frequently on project teams and project-based organizing to perform interdependent tasks, so employees transition more often across project teams - and firms - in their boundary-less careers. These changes call for a better understanding of whether the reactivation of past social ties is likely to help or hinder the job performance of new employees, especially those engaged in highly interdependent tasks. The object of this study is to theorize and empirically test the mechanisms by which the reactivation of a particular social tie - shared work experience - may impact new members' performance. Using a social networks lens to study new members' organizational entries, this study not only contributes to the recent fast-growing literature on the reactivation of social ties, but also to studies on new members' performance, and has considerable relevance for enhancing an organization's performance through the better management of its expert workers' human and social capital. / Thesis (PhD) — Boston College, 2013. / Submitted to: Boston College. Carroll School of Management. / Discipline: Management and Organization.

No description available.

This thesis models the importance of reputation in indefinite interactions using experimental and empirical evidence. In modern society reputation plays an important role in many everyday life situations. I analyse the resulting incentive structures through microeconomic modelling and econometric regression analysis. Thus, I gauge the impact of reputational considerations on individual decision making.

Opinion dynamics in a society of interacting agents may lead to consensus or to the coexistence of different opinions. The interplay between social network change and opinion formation is complex, because the agents, their social interactions and the changing social structure over time, are themselves complex. DeGroot proposed a prescriptive model for achieving consensus, where agents revise their opinions at each time step by taking a weighted average of the opinions of neighbours. This thesis contains three main contributions. First, we introduce a generalisation of the DeGroot model and examine the long-time behaviour of the model, with and without insistent agents. Second, we consider opinion formation on networks which are themselves dynamic, where the dynamics may be completely random or based on homophily and triadic closure. The weights that agents place on the opinions of neighbours are also dynamic, based on a rule where weights decrease with increased difference in opinions. Third, we examine the effect of a sudden, temporary or permanent shift in the opinions of some agents. Two dynamics are considered for the network change over time; random switching (RS) network dynamics, and homophily and triadic closure (HT) network dynamics. We prove that the RS network dynamics enhances consensus formation and network connectivity, compared to the HT network dynamics where we show by simulation that different opinions can persist. We investigate the in uence of the presence of a minority of insistent agents and prove that for a connected static network, insistent agents with the same opinion in uence the final opinions to converge to their own opinion, thus leading to consensus. In contrast, lack of consensus persists when insistent agents have different opinions. This conclusion also holds for the RS network dynamics model. However, for the HT network dynamics model, coexistence of different opinions can persist even when insistent agents have the same opinion. This finding regarding the HT dynamics is of particular interest as it relates to observations in the real-world. We also investigate the in uence of a sudden shift in the opinions of some agents on the outcome of final opinions. The case of either a temporary shift in opinions or a permanent shift in opinions is examined. Additionally, the in uence of the time of the introduction of a shift, the number and the network positions of initial recipients of the shift in opinions is investigated. The overall effect of an opinion shift is measured by its in uence on the stabilisation time of the final opinions.

Zhu, Yi. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2011. / Includes bibliographical references (p. 136-148). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgement --- p.vi / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Overview --- p.1 / Chapter 1.2 --- Motivations of Detecting Commercial Intention --- p.4 / Chapter 1.3 --- Problem Definition for Commercial Intention Detection --- p.6 / Chapter 1.4 --- Contributions --- p.8 / Chapter 1.5 --- Thesis Organization --- p.9 / Chapter 2 --- Literature Review --- p.12 / Chapter 2.1 --- Twitter and Tweets Analysis --- p.13 / Chapter 2.2 --- Intention Detection --- p.17 / Chapter 2.2.1 --- User Intention Mining --- p.17 / Chapter 2.2.2 --- Commercial Intention Mining --- p.18 / Chapter 2.3 --- Similar Task: Opinion Mining --- p.18 / Chapter 2.4 --- NLP Techniques for Commercial Intention Detection --- p.20 / Chapter 2.4.1 --- Words Semantic Similarity --- p.21 / Chapter 2.4.2 --- Short Text Similarity --- p.25 / Chapter 2.5 --- Hierarchical Classification --- p.26 / Chapter 2.5.1 --- Hierarchical Classifiers Overview --- p.26 / Chapter 2.5.2 --- Construction of Hierarchy --- p.27 / Chapter 2.5.3 --- Taxonomy of Hierarchical Classification --- p.28 / Chapter 3 --- System Overview --- p.31 / Chapter 3.1 --- Feasibility of Commercial Intention Detection --- p.31 / Chapter 3.2 --- System Design and Architecture --- p.33 / Chapter 3.3 --- Components of READ-MIND --- p.35 / Chapter 3.3.1 --- Preprocessing --- p.35 / Chapter 3.3.2 --- Centroid Word Locator --- p.37 / Chapter 3.3.3 --- Commercial Intention Detector --- p.38 / Chapter 3.3.4 --- Tweet Classifier --- p.40 / Chapter 3.3.5 --- Advertisement Mapping --- p.41 / Chapter 3.4 --- System Work Flow --- p.42 / Chapter 3.4.1 --- System Dataflow and Controlflow --- p.42 / Chapter 3.4.2 --- User Interface --- p.42 / Chapter 3.5 --- System Speed Up --- p.43 / Chapter 3.6 --- Summary --- p.45 / Chapter 4 --- Natural Language Processing on Tweets --- p.46 / Chapter 4.1 --- NLP Techniques in READ-MIND --- p.46 / Chapter 4.2 --- Centroid Word Locator --- p.47 / Chapter 4.2.1 --- Centroid Word --- p.47 / Chapter 4.2.2 --- Locating Centroid Word --- p.48 / Chapter 4.2.3 --- Centroid Word Pair --- p.50 / Chapter 4.2.4 --- Locating Centroid Word Pair --- p.54 / Chapter 4.3 --- Semantic Relatedness Between Tweets --- p.59 / Chapter 4.3.1 --- Relatedness with a Words Set --- p.60 / Chapter 4.3.2 --- Relatedness between Tweets --- p.62 / Chapter 4.3.3 --- Words Similarity --- p.63 / Chapter 4.4 --- Summary --- p.65 / Chapter 5 --- Tweets Classification --- p.66 / Chapter 5.1 --- Two Stages of Tweets Classification --- p.66 / Chapter 5.2 --- Commercial Intention Detector --- p.68 / Chapter 5.2.1 --- Intuitive Method --- p.68 / Chapter 5.2.2 --- Binary Classification --- p.70 / Chapter 5.3 --- Tweet Categorization --- p.72 / Chapter 5.3.1 --- Build Hierarchical Classifier --- p.73 / Chapter 5.3.2 --- Hierarchical Classification --- p.81 / Chapter 5.4 --- Summary --- p.83 / Chapter 6 --- Empirical Study --- p.84 / Chapter 6.1 --- Objective of Empirical Study --- p.84 / Chapter 6.2 --- Experiment Setup and Evaluation Methodology --- p.85 / Chapter 6.2.1 --- Simulation Environment --- p.85 / Chapter 6.2.2 --- Tweets Data Set --- p.86 / Chapter 6.2.3 --- Labeling Process --- p.87 / Chapter 6.2.4 --- Evaluation Methodology --- p.88 / Chapter 6.3 --- Compare Algorithms in Components --- p.90 / Chapter 6.3.1 --- Centroid Word VS. Centroid Word Pair --- p.91 / Chapter 6.3.2 --- Semantic Similarity Comparison --- p.92 / Chapter 6.3.3 --- Methods in Commercial Intention Detector --- p.93 / Chapter 6.3.4 --- Structure of Hierarchy --- p.94 / Chapter 6.3.5 --- Training Source of Tweets Classifier --- p.95 / Chapter 6.3.6 --- Summary --- p.96 / Chapter 6.4 --- Parameter Settings Comparison --- p.97 / Chapter 6.4.1 --- Impact of Varying Parameters --- p.97 / Chapter 6.4.2 --- Discussion on Parameter Setting --- p.98 / Chapter 6.5 --- Comparison of READ-MIND and Baseline Method --- p.100 / Chapter 6.6 --- Time Cost Analysis --- p.101 / Chapter 6.6.1 --- Time Cost to Process Tweets --- p.101 / Chapter 6.6.2 --- Comparison with Baseline --- p.102 / Chapter 6.6.3 --- Analysis on Real-Time Property --- p.103 / Chapter 6.7 --- TCI Categories Comparison --- p.106 / Chapter 6.7.1 --- Results for Different TCIs --- p.106 / Chapter 6.7.2 --- Comparison of Different TCIs --- p.107 / Chapter 6.8 --- Summary --- p.108 / Chapter 7 --- Conclusion --- p.109 / Chapter 7.1 --- Conclusion --- p.109 / Chapter 7.2 --- Future Work --- p.111 / Chapter A --- List of Abbreviations --- p.112 / Chapter B --- List of Symbols --- p.114 / Chapter C --- Proof --- p.117 / Chapter D --- System Work Flow --- p.120 / Chapter E --- Algorithms --- p.123 / Chapter F --- Detailed Experimental Results --- p.129 / Bibliography --- p.136

With the increasing prevalence of large scale online social networks, the field has evolved from studying small scale networks and interactions to massive ones that encompass huge fractions of the world’s population. While many methods focus on techniques at scale applied to a single domain, methods that apply techniques across multiple domains are becoming increasingly important. These methods rely on understanding the complex relationships in the data. In the context of social networks, the big data available allows us to better model and analyze the flow of information within the network. The first part of this thesis discusses methods to more effectively learn and predict in a social network by leveraging information across multiple domains and types of data. We document a method to identify users from their access to content in a network and their click behavior. Even on a macro level, click behavior is often hard to obtain. We describe a technique to predict click behavior using other public information about the social network. Communication within a network inevitably has some bias that can be attributed to individual preferences and quality as well as the underlying structure of the network. The second part of the thesis characterizes the structural bias in a network by modeling the underlying information flow as a commodity of trade.

One of the important issues in research on social support has been the relationship between received support and perceptions of support. While most research has been unable to discover a strong link between these two constructs, three theories have emerged in the literature to explain how such a link might be discovered. One theory states that it is important to study support in the context of a stressful life event. Another suggests that when studying social support it is important to make distinctions between positive and negative interactions. A third approach focuses on specificity issues, predicting that it is important to specify the source, type and timing of support. This study is concerned with widowhood and satisfaction with support. It addresses the questions present in the social support literature by focusing on five hypotheses. The first hypothesis predicts that while there will be more reported support than problems, the effects of the problems will be greater than the effects of the support. The second hypothesis anticipates a stronger link between received and perceived support than other studies have indicated. This is because this study is focusing on a widowhood as a stressful life event and is also differentiating between positive and negative interactions. The third, fourth and fifth hypotheses focus on issues of specificity. The third hypothesis predicts that there will be a difference in satisfaction with family and non-family support. The fourth hypothesis looks at types of support and suggests that different forms of support and problems will affect satisfaction with family differently than satisfaction with non-family. The fifth hypothesis adds the temporal component, anticipating that satisfaction with different types of support and problems from family and non-family will vary over time. Data for this study comes from the first year of a three year longitudinal research project conducted by the Institute on Aging at Portland State University. The sample consists of widows who live in the metropolitan area of Portland, Oregon. Potential respondents were contacted if their names appeared as the surviving spouse on a sample of death certificates. Widows were first contacted by mail, and if they indicated interest they were contacted later by phone. Women who were interested and eligible to participate were divided into three groups depending on length of time widowed. All the data used in this study is the result of one and a half hour long face-to-face interviews with each of the respondents. Received support was measured by asking respondents detailed questions about the kinds of help and problems they received from different network members. Perceived support was measured on a seven point scale which rated how satisfied widows were with their family and their non-family networks. Other important variables have to do with length of time widowed, size of networks, and frequency of contact with family and non-family network members. Despite the overall prediction, that the amounts of support received will affect a person's satisfaction with support, the data only partially supported the five hypotheses. All the links between support, problems and satisfaction were in the non-family network. The only time that received support seemed to be significant was when examining non-family instrumental support among the most recent widows. Problematic interactions had increasingly stronger effects on satisfaction as the amount of time widowed increased.

acase@tulane.edu / Understanding the causes and consequences of plumage ornamentation in birds has long been of interest to evolutionary ecologists. Although conspicuous plumage is often present year-round, and is therefore likely to be under selection throughout the annual cycle, it has been much better-studied during breeding than non-breeding seasons. Signaling function of ornaments may be distinct during the non-breeding season because of seasonal differences in ecological and evolutionary pressures. The red-backed fairywren is a small, insectivorous passerine bird from Australia in which males have variable plumage ornamentation: some males develop red/black plumage months before breeding, others do so immediately before breeding, and others retain cryptic brown plumage year-round. My thesis investigated the causes and consequences of this variation, which a focus on non-breeding season dynamics. In my first chapter, I characterized the hormonal basis - or lack thereof - for ornamented plumage for early-molting males, finding that androgen levels in both males and females regardless of molt status were similar during the non-breeding season. By expressing plumage ornamentation in the absence of elevated androgen levels, ornamented males may avoid immunological or other costs associated with androgens. In my second chapter, I found that early-molting males did not pay a cost in terms in terms of signal quality relative to males that developed plumage ornamentation later in the year, when resources might be more available. Instead, males that developed ornamented plumage months before breeding had putatively higher quality ornamentation during the breeding season, and I identified adventitious molt in these males as a novel mechanism for plumage modification. In my third chapter, I showed that ornamented males ranged over larger areas and had more social connections that unornamented males. These males also displayed to females months before the start of the breeding season, suggesting that space use during the non-breeding season may be in part driven by opportunities to interact with future reproductive partners. This work advances our understanding of the production mechanisms, signaling function, and behavioral correlates of plumage ornamentation during non-breeding periods. / 1 / Samantha Lantz