Bakgrund: Under lång tid har den traditionella tolkningen varit att marknadspremier och högre avkastning på aktiemarknaden är kopplat till risk. Även den mest använda prissättningsmodellen idag, Capital Asset Pricing Model, bygger på detta antagande. I en artikel skriven av Ibbotson och Idzorek (2014) utmanas dock risk som den viktigaste faktorn bakom premier och avkastning. Artikeln innehåller stöd för att relativt hög avkastning har kunnat uppnås på den amerikanska marknaden genom att investera i portföljer med aktier som föregående år varit relativt opopulära. Den höga avkastningen genererades dessutom ofta till relativt låg risk. Intresse finns därmed att analysera effekten av aktiers popularitet även på den svenska marknaden. Syfte: Studiens syfte är att identifiera och analysera effekten av aktiers popularitet på avkastning och risk. Genomförande: I denna kvantitativa studie har aktieomsättningshastighet och aktiestorlek utgjort approximationer för popularitet. Studien har genomförts via utvärdering av avkastning och risk i aktieportföljer uppdelade utifrån variablerna aktieomsättningshastighet och storlek. Vidare har sambandet mellan popularitet och avkastning undersöks via linjär regressionsanalys. Studien har både undersökt effekten av föregående års popularitet, samt effekten av popularitet samma år. Slutsats: Studien visar ingen entydig effekt för aktiers popularitet föregående år på avkastning eller risk, när olika approximationer för popularitetsmått studeras och jämförs. Studien kan konstatera att det inte finns något samband mellan föregående års popularitet och avkastning. Däremot finns det ett positivt samband mellan popularitet och avkastning de år aktiernas popularitet uppmätts, när aktieomsättningshastighet används som approximation. Dessutom kan studien fastslå stöd för aktieomsättningshastighet som ett bra mått på aktiers popularitet. / Background: Over the past few decades it has been generally accepted that market premiums come with an associated level of risk. Even the most widely used pricing model today, CAPM, leans on this assumption. In an article written by Ibbotson and Idzorek (2014) this assumption is challenged as the main driver of market premiums and returns. The article contains evidence that relatively high returns have been earned through buying less  popular stocks on the U.S. stock market. Surprisingly the risk-return dimension exhibited an inverse relationship. This evidence from the U.S. stock market motivates us to investigate to what extent this effect can also be seen on the Swedish stock market. Aim: The aim of this thesis is to identify and analyze the effect of a stock`s popularity on the risk and return. Completion: In this quantitative study, share turnover and market capitalization have been used as approximations for popularity. The effects of stocks popularity on risk and return have been are examined by evaluating the performance of portfolios when categorizing the stocks by share turnover and market capitalization. The statistical relationship between popularity and return is analyzed using regression analysis. This study has both studied the effect of last year's popularity, as well as the effect of the popularity of the same year. Conclusion: When various approximations for the popularity dimension are studied and compared, this study shows no marked effect of stock`s popularity from the previous year on risk and return. The study finds no statistically significant relationship between the previous year ́s popularity and return. However, there is a positive statistically correlation between popularity and return when measured during the same year as when the popularity was measured. In addition, the results establish evidence for the stock turnover as a good measure of popularity.

abstract: Crises or large-scale emergencies such as earthquakes and hurricanes cause massive damage to lives and property. Crisis response is an essential task to mitigate the impact of a crisis. An effective response to a crisis necessitates information gathering and analysis. Traditionally, this process has been restricted to the information collected by first responders on the ground in the affected region or by official agencies such as local governments involved in the response. However, the ubiquity of mobile devices has empowered people to publish information during a crisis through social media, such as the damage reports from a hurricane. Social media has thus emerged as an important channel of information which can be leveraged to improve crisis response. Twitter is a popular medium which has been employed in recent crises. However, it presents new challenges: the data is noisy and uncurated, and it has high volume and high velocity. In this work, I study four key problems in the use of social media for crisis response: effective monitoring and analysis of high volume crisis tweets, detecting crisis events automatically in streaming data, identifying users who can be followed to effectively monitor crisis, and finally understanding user behavior during crisis to detect tweets inside crisis regions. To address these problems I propose two systems which assist disaster responders or analysts to collaboratively collect tweets related to crisis and analyze it using visual analytics to identify interesting regions, topics, and users involved in disaster response. I present a novel approach to detecting crisis events automatically in noisy, high volume Twitter streams. I also investigate and introduce novel methods to tackle information overload through the identification of information leaders in information diffusion who can be followed for efficient crisis monitoring and identification of messages originating from crisis regions using user behavior analysis. / Dissertation/Thesis / Doctoral Dissertation Computer Science 2015

abstract: The increasing popularity of Twitter renders improved trustworthiness and relevance assessment of tweets much more important for search. However, given the limitations on the size of tweets, it is hard to extract measures for ranking from the tweet's content alone. I propose a method of ranking tweets by generating a reputation score for each tweet that is based not just on content, but also additional information from the Twitter ecosystem that consists of users, tweets, and the web pages that tweets link to. This information is obtained by modeling the Twitter ecosystem as a three-layer graph. The reputation score is used to power two novel methods of ranking tweets by propagating the reputation over an agreement graph based on tweets' content similarity. Additionally, I show how the agreement graph helps counter tweet spam. An evaluation of my method on 16~million tweets from the TREC 2011 Microblog Dataset shows that it doubles the precision over baseline Twitter Search and achieves higher precision than current state of the art method. I present a detailed internal empirical evaluation of RAProp in comparison to several alternative approaches proposed by me, as well as external evaluation in comparison to the current state of the art method. / Dissertation/Thesis / M.S. Computer Science 2013

This paper investigates presence of the superstar effect among users of the Czech fantasy league during seasons between years 2010 and 2013. Both main views are taken into consideration, namely Rosen's and also Adler's theory. Using regression models employing the method of least squares for panel data I show a positive superstar effect according to Rosen's theory in case when the star players are defined by their ability to score great amounts of goals. The presence of a positive superstar effect according to Adler's theory is not apparent in the models, which can be caused by the authors of the fantasy league overestimating the value of the star players beside their performance.

The purpose which guided the writer in the conducting of this study was a desire to determine the existence of any relationships which might be perceptible among such traits and considerations as socio-economic status, popularity at school, general scholastic achievement, and traits of personality as they were found in a group of fifth-grade pupils enrolled in the public school of Sanger, Texas.

De nos jours, la popularité des réseaux sociaux (RS) est en constante progression. En effet, de plus en plus d’utilisateurs interagissent dans le monde virtuel, soit en y exprimant des opinions, en partageant des expériences, en réagissant aux avis d’autrui ou encore en échangeant des idées, en fonction de leurs qualités : influents, populaires, dignes de confiance, etc.. Dans la littérature, l’influence a fait l'objet d'une attention particulière ces dernières années. En effet, de nombreux domaines, dont l’Analyse des Réseaux Sociaux (ARS) et les systèmes de recommandation ont étudié l’influence, sa détection, la propagation de son effet et sa mesure. Ainsi, des modèles d'identification et d'estimation de l'influence sont aujourd'hui largement utilisés dans de nombreuses applications dédiées au marketing, aux campagnes politiques/sociales, etc. De plus, les interactions entre utilisateurs indiquent non seulement l’influence mais aussi la confiance, la popularité ou la réputation. Cependant, ces notions sont encore vaguement définies et il n'existe pas de consensus dans la communauté ARS. Définir, distinguer et mesurer la force de ces relations entre les utilisateurs posent également de nombreux défis, à la fois théoriques et pratiques, qui restent à explorer. La modélisation de l’influence pose de multiples défis et les méthodes actuelles de découverte et d’évaluation n’explorent pas encore pleinement les différents types d’interactions et ne sont en général pas applicables à plusieurs RS. En outre, la prise en compte de la dimension temporelle dans le modèle d’influence est importante, difficile et nécessite un examen plus approfondi. Enfin, l’exploration de liens possibles entre des notions, telles que l’influence et la réputation, reste un sujet ouvert. Dans cette thèse, nous nous focalisons sur les quatre concepts qualifiant les utilisateurs : influence, réputation, confiance et popularité, pour la modélisation de l'influence. Nous analysons les travaux existants utilisant ces notions et comparons leurs différentes interprétations. Par cette analyse, nous mettons en avant les caractéristiques essentielles que ces concepts devraient inclure, et nous en effectuons une analyse comparative. Cela nous permet d'établir une classification globale des différentes interprétations des notions selon leur niveau d'abstraction et leurs divergences ; cela constitue la première, contribution de cette thèse. En conséquence, nous proposons un modèle théorique de l'influence ainsi qu'une ontologie associée décrivant ce concept. Nous présentons également une variante de l'influence, inexplorée à ce jour dans le domaine de l’ARS, la micro-influence. Celle-ci cible un phénomène nouveau dans les RS que sont les utilisateurs avec une faible audience, mais fortement impliqués ; ces derniers apparaissent en effet comme ayant un impact fort malgré tout. En s'appuyant sur ces définitions, nous proposons ensuite un modèle pratique dénommé ARIM (Action-Reaction Influence Model). Ce modèle considère le type, la qualité, la quantité et la fréquence des actions réalisées par les utilisateurs, et ce en étant compatible avec différents RS. Nous abordons également la quantification de l'influence au cours du temps et la représentation de ses effets de causalité. Pour cela, nous considérons un type spécifique de RS: les réseaux de citations, particulièrement sensibles au temps. Ainsi, nous proposons un modèle, TiDIE (Time Dependent Influence Estimation), qui détermine l'influence, sur une période de temps, entre les communautés de ces réseaux. Enfin, nous combinons l’influence et la réputation avec le modèle TiDIE, afin d’étudier les dépendances entre elles. Nous proposons une méthode de transition, ReTiDIE, utilisant l’influence pour obtenir la réputation. Pour chacune de nos approches, des expérimentations ont été menées sur des jeux de données réels et ont montré la pertinence de nos méthodes / Online social networks are constantly growing in popularity. They enable users to interact with one another and shifting their relations to the virtual world. Users utilize social media platforms as a mean for a rich variety of activities. Indeed, users are able to express their opinions, share experiences, react to other users' views and exchange ideas. Such online human interactions take place within a dynamic hierarchy where we can observe and distinguish many qualities related to relations between users, concerning influential, trusted or popular individuals. In particular, influence within Social Networks (SN) has been a recent focus in the literature. Many domains, such as recommender systems or Social Network Analysis (SNA), measure and exploit users’ influence. Therefore, models discovering and estimating influence are important for current research and are useful in various disciplines, such as marketing, political and social campaigns, recommendations and others. Interestingly, interactions between users can not only indicate influence but also involve trust, popularity or reputation of users. However, all these notions are still vaguely defined and not meeting the consensus in the SNA community. Defining, distinguishing and measuring the strength of those relations between the users are also posing numerous challenges, on theoretical and practical ground, and are yet to be explored. Modelization of influence poses multiple challenges. In particular, current state-of-the-art methods of influence discovery and evaluation still do not fully explore users’ actions of various types, and are not adaptive enough for using different SN. Furthermore, adopting the time aspect into influence model is important, challenging and in need of further examination part of the research. Finally, exploring possible connections and links between coinciding notions, like influence and reputation, remains to be performed.In this thesis, we focus on the qualities of users connected to four important concepts: influence, reputation, trust, and popularity, in the scope of SNA for influence modeling. We analyze existing works utilizing these notions and we compare and contrast their interpretations. Consequently, we emphasize the most important features that these concepts should include and we make a comparative analysis of them. Accordingly, we present a global classification of the notions concerning their abstract level and distinction of the terms from one another, which is a first and required contribution of the thesis. Consequently, we then propose a theoretical model of influence and present influence-related ontology. We also present a distinction of notion not yet explored in SNA discipline -- micro-influence, which targets new phenomena of users with a small but highly involved audience, who are observed to be still highly impactful. Basing on the definitions of the concepts, we propose a practical model, called Action-Reaction Influence Model (ARIM). This model considers type, quality, quantity, and frequency of actions performed by users in SN, and is adaptive to different SN types. We also focus on the quantification of influence over time and representation of influence causal effect. In order to do that, we focus on a particular SN with a specific characteristic - citation network. Indeed, citation networks are particularly time sensitive. Accordingly, we propose Time Dependent Influence Estimation (TiDIE), a model for determining influence during a particular time period between communities within time-dependent citation networks. Finally, we also combine two of the abovementioned notions, influence and reputation, in order to investigate the dependencies between them. In particular, we propose a transition method, ReTiDIE, that uses influence for predicting the reputation. For each of the proposed approaches, experiments have been conducted on real-world datasets and demonstrate the suitability of the methods

The diploma thesis is focused on the problems of children readership and its main aim is to reveal the influence of the literary work that is very popular among readers on the interpretation of concrete literary text. By the means of research it tries to detect how the work Harry Potter by J. K. Rowling and its popularity affect the perception of literary text with respect to pupils of the 6th grade in primary school and students of 1st year in secondary school. The defined aim is studied mainly through reader's perception of literary characters in the concrete literary text; the main focus is aimed at perception of character hierarchy, creation of relationships with characters and perception of the main character of the work Harry Potter. Thanks to these findings it is possible to say whether the children readers are able to perceive the concrete literary text independently on the work by J. K. Rowling and its popularity or whether they are constantly under its influence and are not able to free themselves from its dominance. According to final discoveries it is possible to approach the work within the context of literary classes and overall literary education. Summarizing the results, the research has revealed that the influence of the phenomenon called Harry Potter and its popularity is quite...

<div>In the past decade, Big Data analysis has become a central part of many industries including entertainment, social networking, and online commerce. MapReduce, pioneered by Google, is a popular programming model for Big Data analysis, famous for its easy programmability due to automatic data partitioning, fault tolerance, and high performance. Majority of MapReduce workloads are summarizations, where the final output is a per-key ``reduced" version of the input, highlighting a shared property of each key in the input dataset.</div><div><br></div><div>While MapReduce was originally proposed for massive data analyses on networked clusters, the model is also applicable to datasets small enough to be analyzed on a single server. In this single-server context the intermediate tuple state generated by mappers is saved to memory, and only after all Map tasks have finished are reducers allowed to process it. This Map-then-Reduce sequential mode of execution leads to distant reuse of the intermediate state, resulting in poor locality for memory accesses. In addition the size of the intermediate state is often too large to fit in the on-chip caches, leading to numerous cache misses as the state grows during execution, further degrading performance. It is well known, however, that many large datasets used in these workloads possess a Zipfian/Power Law skew, where a minority of keys (e.g., 10\%) appear in a majority of tuples/records (e.g., 70\%). </div><div><br></div><div>I propose ZipThru, a novel MapReduce software architecture that exploits this skew to keep the tuples for the popular keys on-chip, processing them on the fly and thus improving reuse of their intermediate state and curtailing off-chip misses. ZipThru achieves this using four key mechanisms: 1) Concurrent execution of both Map and Reduce phases; 2) Holding only the small, reduced state of the minority of popular keys on-chip during execution; 3) Using a lookup table built from pre-processing a subset of the input to distinguish between popular and unpopular keys; and 4) Load balancing the concurrently executing Map and Reduce phases to efficiently share on-chip resources. </div><div><br></div><div>Evaluations using Phoenix, a shared-memory MapReduce implementation, on 16- and 32-core servers reveal that ZipThru incurs 72\% fewer cache misses on average over traditional MapReduce while achieving average speedups of 2.75x and 1.73x on both machines respectively.</div>

I Sverige hamnar ungdomar mellan 18 och 19 år i en gråzon angående laglighet och alkohol. Tidigare forskning finns i mängder om ungdomar och alkohol, men en lucka finns gällande Sveriges gråzon. Denna studie hade som syfte att besvara hur relationen mellan ungdomars alkoholkonsumtion och popularitet, grupptryck, föräldrars alkoholsyn samt ortstorlek såg ut. Enkäten besvarades av 128 gymnasieungdomar i både mindre och större orter i Sverige. Frågorna gällande alkohol baserades på IQ:s alkoholprofil, och frågorna om popularitet och grupptryck konstruerades ursprungligen av Santor et al (2000). Materialet analyserades i SPSS genom en multipel regressionsanalys och Pearsonkorrelationer. Resultaten visade främst att grupptryck var den starkaste prediktorn för ökad alkoholkonsumtion. Vidare visade det sig att strikta föräldrar genererade lägre alkoholskonsumtion, och att mindre orter innebar mer olagligt drickande. Studien kan bidra till ökad förståelse för ungdomars alkoholkultur samt kan vara till nytta för skolhälsan.

My dissertation consists of two essays that explore important aspects of empirical corporate finance, specifically the importance of political factors and public attention that come in to play in the granting of post-disaster loans. The first paper, “Natural disasters, public attention, and disaster lending”, examines how public attention (as measured by a Google search metric that I constructed) to local natural disasters affects firms’ access to debt. I hypothesize that the lending behavior of creditors in the aftermath of a natural disaster would be strongly influenced by two factors: (1) direct governmental pressure on local and foreign banks, and (2) indirect pressure from local community sentiment. National governments are influenced by public attention around the local natural disaster. They also use the degree of public attention to pressure private banks and state banks to make disaster loans to firms affected by the natural disasters. I posit that the influence and effectiveness of the governmental pressure would be a function of the degree to which the embedding economy has state-owned banks and nationalized banks. Governmental pressure would be limited in its impact in economies that are private (as in the United States or the United Kingdom). The empirical investigation in my first paper will make use of two novel multidimensional cognitive indices using machine learning. The first index (natural disaster intensity index) captures the intensity of the natural disaster. The second index measures the degree of public attention to natural disasters in the local community and is constructed by using Google Trends search data (web searches, image searches, online news searches, and YouTube video searches). Using firm-level and natural disaster data from 30 countries, I document that firms are able to borrow more when there is a heightened public attention to the natural disaster. I also find that different types of media searches (i.e., web searches, online news searches, image searches and YouTube video searches) have differential impact on public attention, and hence, on incremental borrowing by affected firms. I examine the change in debt likelihood as a function of the proportion of image searches (i.e., relative importance of image searches divided by the total of all four types of searches). Here, I observe a nonlinear relationship between the increase in debt likelihood and the proportion of image searches. The increase in debt likelihood has an “inverted U-shaped” relationship with the degree of image searches. I also find similar relationship between the increase in debt likelihood and the proportion of online news searches and web searches. The response of debt likelihood to public attention is higher in countries with a higher historical vulnerability to natural disasters. The response of debt likelihood to public attention is higher following earthquakes and wildfires. I also document an increase in debt likelihood following disasters to which there is heightened public attention in economies with a smaller fraction of state-owned banks. This relationship also obtains in economies with a smaller fraction of foreign banks in the banking sector. This paper addresses important issues of access to debt financing for firms affected by natural disasters. I construct various indices of the degree of community attention and use them as proxies for the importance of political factors and governmental pressure that can influence the change in leverage following the natural disaster. I use novel metrics of public attention based on big data and media search using machine learning for firms around the world. Firms affected by the natural disaster are often at the mercy of access to finance from the relief efforts of the local government and the local banking sector. The availability of disaster loans may have dramatic and long-run effects on the ability of the community to cope with the disaster. Lack of access to capital in such situations (including Covid-19) is an important societal issue, affecting corporate bankruptcies and unemployment. The issue is how the private sector will react to natural disasters with or without government support. This paper provides a novel and behavioral explanation for disaster financing and examines several predictions using novel data and novel metrics to measure the intensity of community sentiment and attention. The second paper, “Polls, Politics and SBA Disaster Loans”, examines the effect of certain important political factors (e.g., the current national popularity of the incumbent U.S. President) on the federal disaster relief effort through the SBA (Small Business Administration)’s disaster loan program. Following natural disasters, there often is a staggering amount of economic damage and even loss of life. A call for government intervention usually follows. In this paper, I use different types of presidencies (i.e., the environmental presidency, the semi-environmental presidency, the pandering presidency, and the classic presidency) to explain the expected impact of current presidential popularity on the willingness of the incumbent U.S. President to authorize federal disaster relief. I also study the influence of the presidential popularity and various related political factors on the intensity of the relief approved and administered through the SBA disaster loan programs. This paper consists of two parts. The first part investigates the impact of the current presidential popularity on the willingness of the incumbent U.S. President to authorize federal disaster relief. Using a unique sample of 1,118 presidential disaster declaration requests from 1991 to 2020, I document an inverted U-shaped effect of the current presidential popularity on the likelihood of a presidential authorization for federal disaster relief. I hypothesize that these results are consistent with the prediction of the semi-environmental presidency model. When the current presidential popularity of the incumbent U.S. President is below 50%, the popularity benefits of using a generous federal disaster relief is important and explains the positive relationship between her popularity and the likelihood of approval. The U.S. President acts like an environmental U.S. President. However, if presidential popularity is greater than 50%, the incumbent U.S. President will be more cautious about authorizing federal disaster relief since the opportunity cost of foregoing important non-environmental related policy initiatives may be higher than the benefits of approving federal disaster relief. The incumbent U.S. President may also supplement the powers granted to her in the U.S. Constitution with the acquired informal powers when her current popularity is higher than 50% in order to realize her own non-environmental related political agenda more easily. In this case, an increase in the U.S. President’s current level of popularity would lead to a decline in the likelihood of her approving federal disaster relief, and they would not be acting as an environmental U.S. President. The second part of this paper investigates how the personal popularity of the incumbent U.S. President impacts the allocation of federal disaster relief to affected counties through the SBA following the authorization of federal disaster relief. I document that the SBA will approve larger amount of disaster loans to disaster-affected households, businesses, and non-profit organizations when the current popularity of the incumbent U.S. President increases. I find that this result is amplified when the incumbent president is (1) a Republican, (2) a second-term president, and (3) not contesting an election in that year. The main findings are robust to different measures of presidential popularity and various estimation methodologies. My contributions in this paper highlight a new venue for politics in climate change in the area of disaster relief. I explore how current public standing of the incumbent U.S. President impacts the disaster relief effort using the SBA disaster loan program. I believe that this is an important area of the interaction of politics and climate finance. Natural disasters and responses to it have become an important topic of the study of climate change, given the increasing frequency and severity of disasters arising from climate change. The politics involving the current pandemic and relief efforts has put this topic in prominent relief (See COVID-19 crisis). / Business Administration/Finance