Interaction event times observed on a social network provide valuable information for social scientists to gain insight into complex social dynamics that are challenging to understand. However, it can be difficult to accurately represent the heterogeneity in the data and to model the dependence structure in the network system. This requires flexible models that can capture the complicated dynamics and complex patterns. Point process models offer an elegant framework for modeling event time data. This dissertation concentrates on developing point process models and related diagnostic tools, with a real data application involving an animal behavior network. In this dissertation, we first propose a Markov-modulated Hawkes process (MMHP) model to capture the sporadic and bursty patterns often observed in event time data. A Bayesian inference procedure is developed to evaluate the likelihood by using a variational approximation and the forward-backward algorithm. The validity of the proposed model and associated estimation algorithms is demonstrated using synthetic data and the animal behavior data. Facilitated by the power of the MMHP model, we construct network point process models that can capture a social hierarchy structure by embedding nodes in a latent space that can represent the underlying social ranks. Our model provides a ranking method for social hierarchy studies and describes the dynamics of social hierarchy formation from a novel perspective – taking advantage of the detailed information available in event time data. We show that the network point process models appropriately captures the temporal dynamics and heterogeneity in the network event time data, by providing meaningful inferred rankings and by calibrating the accuracy of predictions with relevant measures of uncertainty. In addition to developing a sensible and flexible model for network event time data, the last part of this dissertation provides essential tools for diagnosing lack of fit issues for such models. We develop a systematic set of diagnostic tools and visualizations for point process models fitted to data in the dynamic network setting. By inspecting the structure of the residual process and Pearson residual on the network, we can validate whether a model adequately captures the temporal and network dependence structures in the observed data.

Thesis (MPhil)--Stellenbosch University, 2001. / ENGLISH ABSTRACT: It is widely accepted that the Internet has become a valuable resource for social scientists, not just for the purpose of information exchange; via e-mail, discussion groups and electronic journals, but also as a medium for data collection. Its global nature gives a researcher access to a vast range of individuals located around the world. It also opens up access to difficult to hitherto penetrate study areas, such as sensitive research on deviant behaviour. Further, it is claimed that considerable savings to both research budgets and time frames are made possible with the new technology. It is not surprising, therefore, that a substantial body of research, employing the Internet as the primary means of data collection, already exists. This raises a number of questions as to how the Internet fares as a research tool. Are there any important methodological issues that this new approach raises? Do the traditional research methods suffice? Or are adaptations to existing methods necessary when difficulties are encountered? Does such changes affect the more fundamental question of the research design? These questions are considered in this thesis. To answer them, I consider two different types of empirical research designs. The first,' survey research, is a quantitative, numerical design that traditionally has a high level of control. I consider in detail issues of sampling, including non-response, and questionnaire design. The second design examined, ethnographic research, is qualitative, textual and generally has a low level of researcher control. I address the methods used in cyber ethnography and then discuss the considerable ethical concerns that feature in such research. I conclude that, on the whole, the existing methods can be transferred to Internet research. Indeed, some of the problems faced in traditional research are also considerations in Internet studies and can be overcome by employing similar techniques, such as using incentives to reduce non-response rates. However, a number of new problems emerge, such as the lack of paralinguistic cues, which require adaptations to the existing methods in order to produce results that can be considered valid and reliable. However, I also argue that these necessary adaptations to the methods do not affect the underlying principles found in the research design. / AFRIKAANSE OPSOMMING: Dit word algemeen aanvaar dat die Internet In waardevolle hulpmiddel is vir sosiaalwetenskaplikes; nie net vir die uitruil van inligting deur middel van e-pos, besprekingsforums en elektroniese joernale nie, maar ook as In instrument om inligting te versamel. Die globale aard van die internet gee die navorser toegang to In wye spektrum individue internasionaal. Dit verleen ook toegang tot moeilike navorsingsareas, soos sensitiewe navorsing oor afwykende gedrag. Verder word beweer dat dit aansienlike besparings moontlik kan maak vir beide die navorsingsbegroting en tydraamwerk. Dit is dus nie verbasend dat In substansiële hoeveelheid van navorsing, wat die Internet as die primêre bron van dataversameling gebruik, reeds bestaan nie. Dit laat verskeie vrae ontstaan oor hoe die Internet vaar as In navorsingshulpmiddel. Is daar enige belangrike metodologiese kwessies wat hierdie nuwe metode aanraak? Is die tradisionele metodes voldoende? Of moet daar veranderinge aan die huidige metodes aangebring word wanneer probleme ontstaan? Sal hierdie veranderinge die fundamentele aspekte van navorsingsontwerp beïnvloed? Die vrae saloorweeg word in hierdie tesis. Ek gebruik twee empiriese navorsingsontwerpe om die vrae te beantwoord. Die eerste, steekproefnavorsing, is In kwantitatiewe ontwerp wat tradisioneel In hoë vlak van beheer toon. Ek ondersoek in detail kwessies van steekproewe, insluitend geen respons en vraelysontwerp. Die tweede ontwerp wat ondersoek word, etnografiese navorsing, is kwalitatief, tekstueel en toon in die algemeen In lae vlak van navorser beheer. Ek ondersoek die metodes wat gebruik word in kuberetnografie en bespreek dan die etiese vraagstukke wat hierdie navorsing kenmerk. Ek kom tot die gevolgtrekking dat oor die algemeen die huidige metodes toegepas kan word op Internetnavorsing. Inderdaad kan van die probleme wat ondervind word in tradisionele navorsing ook ondervind word in Internet studies en ook hier kan dit oorkom word deur die gebruik van soortgelyke tegnieke, soos om aansporingsbonusse om geen responskoerse te verminder, hoewel daar nuwe probleme opduik, soos die gebrek aan para-taalkundige wenke. Dit noodsaak veranderinge aan die huidige metodes om resultate te lewer wat geldig en betroubaar is. Ek redeneer egter ook dat hierdie nodige veranderinge aan die metodes nie onderliggende beginsels van navorsingsontwerp verander nie.

Indiana University-Purdue University Indianapolis (IUPUI) / In today's fast paced world, people have become increasingly interested in online communication to facilitate their lives and make it faster. This goes on from simple social interactions to more advanced actions like shopping on the internet. The presence of such activities makes it crucial for people to use their common sense and judgment to process all this information and evaluate what/who they trust and what/whom they do not. This process would have been much easier if the number of people in such networks is really small and manageable. However, there are millions of users who are hooked online every day. This makes the person very overwhelmed with his trusting decision, especially when it comes to interacting with strangers over the internet, and/or buying personal items, especially expensive ones. Therefore, many trust models have been proposed by computer scientists trying to evaluate and manage the trust between users using different techniques and combining many factors. What these computer scientists basically do is coming up with mathematical formulas and models to express trust in online networks and capture its parameters. However, social scientists are the people better trained to deal with concepts related to human behaviors and their cognitive thinking such as trust. Thus, in order for computer scientists to support their ideas and get a better insight about how to direct their research, people like social scientists should contribute. With this in mind, we realized in our group work the importance of such contribution, so we came up with the idea of my research work. In my search, I tried to find how these social scientists think and tackle a dynamic notion like trust, so we can use their findings in order to enhance our work and trust model. Through the chapters, I will discuss an already developed trust model that uses measurement theory in modeling trust. I will refer back to this model and see how other social scientists dealt with some of the issues encountered by the model and its functionality. Some small experiments have been done to show and compare our results with social scientists results for the same matter. One of the most important and controversial points to be discussed from social scientists' point of view is whether trust is transitive or not. Other points to be discussed and supported by social scientists' research include aggregation, reputation, timing effects on trust, reciprocity, and experience effects on trust. Some of these points are classified into trust mapping categories and others are related to trust management or decision making stages. In sum, this work is a multidisciplinary study of trust whose overall goal is to enhance our work and results, as computer scientists.