Thesis (Ph. D.)--University of Hong Kong, 1995. / Includes bibliographical references (leaf 408-428).

M.A. (Social Work) / Please refer to full text to view abstract

No description available.

Social networks, such as Twitter and Facebook, enable the wide spread of information through users’ influence on each other. These networks are very useful for marketing purposes. For example, free samples of a product can be given to a few influencers (seed nodes), with the hope that they will convince their friends to buy it. One way to formalize marketers’ objective is through the influence maximization problem, which is to find the best seed nodes to influence under a fixed budget so that the number of people who get influenced in the end is maximized. Recent solutions to influence maximization rely on the knowledge of the influence probability of every social network user. This is the probability that a user influences another one, and can be obtained by using users’ history of influencing others (called action logs). However, this information is not always available. We propose a novel Online Influence Maximization (OIM) framework, showing that it is possible to maximize influence in a social network in the absence of exact information about influence probabilities. In our OIM framework, we investigate an Explore-Exploit (EE) strategy, which could run any one of the existing influence maximization algorithms to select the seed nodes using the current influence probability estimation (exploit), or the confidence bound of the estimation (explore). We then start the influence campaign using the seed nodes, and consider users’ immediate feedback to the campaign to further decide which seed nodes to influence next. Influence probabilities are modeled as random variables and their probability distributions are updated as we get feedback. In essence, we perform influence maximization and learning of influence probabilities at the same time. We further develop an incremental algorithm that can significantly reduce the overhead of handling users’ feedback information. We validate the e↵ectiveness and efficiency of our OIM framework on large real-world datasets. / published_or_final_version / Computer Science / Master / Master of Philosophy

<p>American family relations are formally defined through marriage and descent but these formal distinctions are inadequate to capture the diversity of contemporary family life. Recent demographic trends have led to a diversification of family structures. Alternative, and less institutionalized ties like co-residence and informal partnerships bind an increasing number of families. Clearly defined cultural models do not yet exist for these new relationships. During these demographic changes the cultural dominance of the single breadwinner model has been challenged by women's mass entry into the labor market. New models of fatherhood have begun to emerge and conventional parenting roles may be carried out in diverse ways. A new method is needed to capture the relational processes of new family forms and the heterogeneity of conventional ones. </p><p>I argue families' formal structures can be classified by the things their members do, and the time they share with each other. Network methods sort family structures into discrete types that capture differences in lived experiences. The distinctions differentiating family structures from each another reveal meaningful information about how families are organized in the contemporary context. The four substantive papers in this dissertation each contribute a different demonstration of this fundamental argument. </p><p>First, the method is developed in a familiar context, using conventional distinctions embedded in kinship terms to move one step beyond traditional analyses of the family. Traditional categorical approaches enumerate traditionally defined relationships. We ask instead how patterns of consanguinity and marriage actually combine in American households, making no assumptions about the importance of any particular relation or individual attribute. </p><p>The three papers that follow are further from the traditional categorical approach. I don't assume that descent and marriage are necessary elements of family relationships. Instead, relationship types are defined by patterns of activities that children do with their potential kin. I apply the method to three waves of time use diaries from the Child Development Supplement of the Panel Study of Income Dynamics. Children's relationships with both traditional and new kin types are heterogeneous, yet structured. Next I develop and test a predictive model of parent-child relationships. The results show that allowing salient relationship features to emerge from time use data is fundamental to understanding how parent-child relationships differ by parents' attributes and household characteristics. </p><p>Finally, I examine how relationship types cohere into families. Children have the same type of family when their families are composed of a similar set of relationship types. The relations within most family types are qualitatively similar to each other - if one relationship is broad (or perfunctory) the others are likely to be as well.</p> / Dissertation

No description available.

Background and aims The social and personal relationships of people with learning disabilities were explored, including the characteristics of their social networks, the extent of social integration and the availability of social support. It was hypothesised that people with learning disabilities would be less socially integrated, have more restricted social networks and more limited. social support than a comparison group of people with physical disabilities. Design and participants A mixed methodology was employed. In the-first part of the study participants were 30 people with learning disabilities, a nominated carer for each of the 30 participants and a comparison group of 17 people with physical disabilities. The second part of the investigation consisted of semi-structured interviews with 6 of the people with learning disabilities. Measures Measures used included-the Life Experiences Checklist, the Circles task, the Social Support Self Report, the Functional Support Inventory and the Social Circles Questionnaire. The author devised a semi-structured interview to assess understanding of different kinds of relationships. Transcripts were analysed using content analysis. Results Levels of integration were better than expected in all areas apart from relationships. Participants reported a mean social network size of 11.7, significantly lower than the comparison group. The networks of people with learning disabilities were largely composed of family or friends with learning disabilities whereas non-disabled friends made up the majority of the network for the comparison group. There were few differences between the groups in terms a of perceived social support. Themes identified from the interview data included the provision of emotional support by friends and betrayal of trust in romantic relationships. Implications The findings indicate that- people with learning disabilities may be functionally but not fully socially integrated within the community, thereby lacking opportunities to experience a wide range- of relationships-_ Directions for future research are suggested.

The social structure of animal societies can be instrumental to the evolution and maintenance of animal behaviour. Animal social networks (ASNs) provide a framework with which to visualise, quantify and analyse animals' social structure. The work in this thesis incorporates two areas of ASN research. The first area is the analysis of sparse group-derived data. Observation of group memberships is a widely used method to uncover social preferences. Here this method is used to probe the social structure of a population of Trinidadian guppies (Poecilia reticulata). The network is analysed to ascertain if genetic relatedness may play a role in governing social structure. The bright colourings of male fish are also analysed to see if colour influences male-male associations. The guppy study provided motivation for an investigation into association indices for group-derived data. Existing indices are evaluated using a simulated dataset and a new index is proposed. The second part of this thesis contributes to a new and exciting trend in ASNs in which complete records of animal associations are obtained enabling temporal network analysis to be used. This is applied to a population of New Caledonian crows (Corvus moneduloides) which are of interest particularly for their ability to manufacture and use tools for foraging. Emulations of information flow through the network are used to assess the network's information flow potential. A network structure in which information can spread rapidly could indicate that crows can potentially learn tool use skills from their peers.

Unhealthy diet, low levels of physical activity, high levels of sedentary behaviour and sleep deprivation are important weight related behaviours that have contributed to the increased prevalence of adolescent obesity. Numerous interventions have been developed to improve weight-related behaviours but they are usually focused on the individual and they ignore the effects of social networks on these behaviours. Much of the research in obesity has explored the role of social networks in promoting health through social influence and selection. However, little research has examined how the structure of social networks and the position of the individuals in the network could condition behaviour association (regardless of the underlying mechanism being social influence or selection) in adolescent friendship networks. Examining social network structure, individual position in the network and how they interact with individual behaviour in friendship networks can assist in better understanding the development and persistence of weight-related behaviours in adolescent friendship networks and provides valuable insight on how to modify these behaviours. Hence, this study aims to examine the role of friendship network properties (density, popularity and centrality) on the association between individual's and friends' weight-related behaviours after reviewing the literature and analysing social network and behavioural survey data. Methods are drawn from a set of analytical tools known as 'Social Network Analysis', which uses friendship nomination data from a complete network (socio-metric), along with reported data on diet, physical activity, sedentary behaviour, and sleep deprivation to investigate how friendship network structure is moderating behaviour association between individuals and their friends in the network. Four schools in Kingdom of Bahrain participated in the study with a total of 673 adolescents between the ages of 11 and 15. Findings suggest that there are associations between adolescents and their friends' in multiple weight-related behaviours. There is also evidence for the moderating role of some network properties on these associations. Findings are gender specific, which has implications for gender-tailored interventions.

之前和社交網絡相關的研究，大多數都非常關注積極正面的用戶鏈接關係；與這些研究不同，我們研究同時含有正面與負面鏈接關係的帶符號社交網絡。具體來講，我們特別關注如何在一個帶符號的社交網絡（該網絡又稱為“目標網絡“）中可信並且有效地去預測鏈接關係的符號為正或是為負，且該網絡中僅有一小部份的鏈接關係符號已知，作為訓練樣本。我們採取遷移學習的機器學習方法，借助於另外一個帶符號社交網絡（該網絡又稱為“源網絡“）中充足的鏈接關係符號信息，從而訓練得到一個有效的鏈接關係分類器；需要注意的是，該 “源網絡同“目標網絡，在鏈接關係樣本和鏈接關係符號的聯合分佈上，並不相同。 / 由於在帶符號社交網絡中沒有事先定義好的屬性向量，我們需要構造一種普適的屬性特徵，從而可以把“源網絡“的拓撲結構信息有效地遷移到“目標網絡“中去。借助於構造好的普適屬性，我們使用了一種類似AdaBoost的遷移學習算法，通過調整訓練樣本的權重，從而可以更好地利用“源網絡“中的樣本信息輔助模型的學習。我們使用兩個真實的大型帶符號社交網絡進行實驗，結果顯示我們的遷移學習算法可以較基準方案，在鏈接關係符號預測的準確度上，提高百分之四十。 / Different from a large body of research on social networks that almost exclusively focused on positive relationships, we study signed social networks with both positive and negative links. Specifically, we focus on how to reliably and effectively predict the signs of links in a signed social network (called a target network), where a very small amount of edge sign information is available as the training data. To train a good classifier, we adopt the transfer learning approach to leverage the abundant edge signs from another signed social network (called a source network) which may have a different joint distribution of the observed instance and the class label. / As there is no predefined feature vector for the edge instances in a signed network, we construct generalizable features that can transfer the topological knowledge from the source network to the target. With the extracted features, we adopt an AdaBoost-like transfer learning algorithm with instance weighting to utilize more useful training instances in the source network for model learning. Experimental results on two real large signed social networks demonstrate that our transfer learning algorithm can improve the prediction accuracy by 40% over baseline schemes. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Ye, Jihang. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2012. / Includes bibliographical references (leaves 51-56). / Abstracts also in Chinese. / Abstract --- p.ii / Acknowledgement --- p.v / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Related Work --- p.8 / Chapter 3 --- Problem Formulation --- p.13 / Chapter 4 --- Feature Construction --- p.16 / Chapter 4.1 --- Explicit Topological Features --- p.17 / Chapter 4.2 --- Latent Topological Feature --- p.19 / Chapter 4.2.1 --- Optimization Algorithm --- p.22 / Chapter 4.2.2 --- Convergence Analysis --- p.23 / Chapter 5 --- Edge Sign Prediction by Transfer Learning --- p.28 / Chapter 5.1 --- Transfer Learning with Instance Weighting --- p.29 / Chapter 5.2 --- Training Loss Analysis --- p.31 / Chapter 6 --- Experimental Evaluation --- p.35 / Chapter 6.1 --- Data Preparation --- p.35 / Chapter 6.2 --- Evaluation of Transfer Learning with InstanceWeighting --- p.37 / Chapter 6.3 --- Effectiveness of Topological Features --- p.41 / Chapter 7 --- Conclusion --- p.45 / Chapter A --- Proof of Theorem 1 --- p.48 / Bibliography --- p.51