Les accords de commerce préférentiels (ACP) sont devenus des instruments importants au travers desquels lesEtats souverains conduisent leurs politiques économiques et façonnent leurs relations politiques internationales.Cette thèse analyse les enjeux transversaux et multidimensionnels des ACP. Dans un premier chapitre, nousmontrons que les domaines de négociation abordés dans les ACP, qui ne relèvent pas du mandat del’organisation mondiale du commerce (OMC), augmentent les échanges commerciaux. De plus, les mécanismesde règlement des différends (en anglais DSM) les plus contraignants favorisent le commerce alors que lesmécanismes moyennement contraignants ne sont pas efficaces. Le second chapitre se tourne vers le cotépolitique des ACP et identifie un canal important au travers duquel les DSM incorporés aux ACP peuvent avoirun impact profond sur la prévention des conflits militaires. Nous montrons que les ACP avec les DSMmoyennement contraignantes empêchent l’escalade des sanctions vers les conflits militaires, tandis que les pluscontraignants ne sont pas efficaces. En plus, les Etats qui sont proche politiquement, économiquement etsocialement, sont plus portés à conclure les DSM les plus contraignants. Le troisième chapitre identifie les ACPselon le niveau de légalisme des clauses d’investissement. Nous montrons que les investisseurs internationauxfont face aux risques d’expropriation liés au régime politique du pays hôte ainsi qu'aux relations diplomatiquesentre leur pays et le pays hôte. Notre analyse montre qu'un régime démocratique ainsi que les bonnes relationsdiplomatiques conditionnent positivement les effets des clauses d’investissement légalement contraignantes.Enfin, le quatrième chapitre se penche sur les clauses environnementales pouvant figurer dans les ACP. Ilidentifie les ACP selon le niveau de légalisme des clauses environnementales. Nous montons que les pays Nord-Nord concluent les clauses environnementales avec le niveau le plus contraignant alors que les paires Nord-Sudsont plus sujettes à adopter les clauses environnementales moyennement contraignantes. / Preferential Trade Agreements (PTAs) have become important instruments through which sovereign statesconduct international economic as well as foreign policies. This thesis analyses multidimensional and crosscuttingissues in PTAs. In the first chapter, we show that policy issues incorporated in PTAs, but outside themandate of World Trade Organization (WTO), have a positive effect on trade. More specifically, highlylegalistic dispute settlement mechanisms (DSM) promote trade liberalization while medium legalistic DSMfeatures do not. The second chapter turns to the political side of PTAs and identifies an important channelthrough which DSMs in PTAs may have profound impact on the prevention of military conflicts. We show thatPTAs with medium level of DSMs prevent the escalation of sanctions to military conflicts, whereas higher leveldoes not. Further, the states that are more politically, economically and socially connected are more prone toform highly legalistic framework in PTAs. The third chapter identifies PTAs according to the level of legalismof investment provisions. Further, we show that international investors face risk of expropriation related to thedomestic political regime of host country as well as the diplomatic relations between home and host countries.Our analysis shows that highly democratic regime as well as the good diplomatic relationship conditionpositively the effects of legally enforced investment provisions. Finally, the fourth chapter identifies PTAsaccording to different levels of legalism of environmental provisions. We show that North-North countries signhighly legalistic environmental provisions whereas North-South country-pairs are more prone to sign mediumlegalistic environmental provisions.

Force deployment and optimal resource allocation has been an area of considerable research interest in conventional warfare. In the modern scenario, with significant advances in technology related to communication and computation, sophisticated decision-making in these situations has become feasible. This has generated renewed interest in formulating decision-making problems in these areas and seeking optimal solutions to them. This thesis addresses one such problem in which the defending forces need to optimally Partition their resources between several attacking forces of differing strengths. The basic model considered for resource allocation is Lanchester attrition models. Lanchester models are deterministic differential equations that model attrition to forces in convict. In this thesis we address a resource allocation problem where the resource allocation is done using different approaches. First, we developed a (2,1) model using the Lanchester square law model for attrition. For this model we assumed that the attacking force consists of two types of forces and the defending force consists of only one type of force. The objective is to optimally partition the defending force against the two attacking forces so as to maximize the surviving defending force strength and to minimize the attacking force strength. The objective function considered in this thesis is the weighted sum of the surviving defending force strength and the destroyed attacking force strength. We considered a resource allocation problem in which allocation of resources are done using four different approaches. The simplest is the case when allocation is done initially and no further action is taken Iv Abstract v (Time Zero Allocation (TZA)). For the TZA allocation scheme, when any of attacking forces gets destroyed, the corresponding defending force which was engaging that attacking force will stop interacting further. This situation rarely happens in reality. Hence to make this scenario more realistic, we considered another allocation scheme in which allocation is followed by redistribution of resources depending on certain decisive events (Time Zero Allocation with Redistribution (TZAR)). In TZA and TZAR schemes, the allocation of defending force is done only at the initial time. Deviating from this assumption, we considered another allocation scheme in which a constant allocation ratio is used continuously over time till the end of the convict (Continuous Constant Allocation (CCA). To account for the redistribution of resources we extended this allocation scheme to the case in which continuous constant allocation is followed by redistribution of the resources (Continuous Constant Allocation with Redistribution (CCAR)). In each of these formulations we define the conditions for an optimal resource partitioning and allocation. We were able to obtain analytical expression for resource partitioning in almost all of these cases. Next, in order to consider situations in which area fire is required, we developed a (2,1) model using Lanchester linear law model for attrition. Here we considered a resource allocation problem in which the resource allocation is done using ideas similar to the square law case. In the Linear law, the resources will get destroyed completely only at infinite time, hence a situation for redistribution of resources does not arise for this law. We considered Time Zero Allocation and Continuous Constant Allocation schemes for this law. We obtained analytical results for the TZA scheme. For the CCA scheme, closed form solutions are difficult to obtain but numerical solutions were obtained. The above schemes were extended to an (n, 1) model for resource allocation using Lanchester square and linear laws. Here the defending forces have to determine an optimal partitioning of available resources to counter attacks from an adversary from n different fronts. For the square law model, we considered both TZA and CCA schemes for resource allocation. As the number of force types increases, the equations becomes much more complicated and the analytical solutions are difficult to obtain. We were able to obtain analytical solutions for some of the situations that occurs during the conflict. For the linear law, we considered only the TZA scheme since, even for the simpler (2,1) model, the analytical solutions are difficult to obtain for the CCA scheme. The resource allocation strategies developed in this thesis contribute to the growing research in the field of conflicts. The thesis concludes with a discussion on some future Extensions of this work.

In this thesis, we have studied the problem of detecting automated Twitter accounts related to the Ukraine conflict using supervised learning. A striking problem with the collected data set is that it was initially lacking a ground truth. Traditionally, supervised learning approaches rely on manual annotation of training sets, but it incurs tedious work and becomes expensive for large and constantly changing collections. We present a novel approach to synthetically generate large amounts of labeled Twitter accounts for detection of automation using a rule-based classifier. It significantly reduces the effort and resources needed and speeds up the process of adapting classifiers to changes in the Twitter-domain. The classifiers were evaluated on a manually annotated test set of 1,000 Twitter accounts. The results show that rule-based classifier by itself achieves a precision of 94.6% and a recall of 52.9%. Furthermore, the results showed that classifiers based on supervised learning could learn from the synthetically generated labels. At best, the these machine learning based classifiers achieved a slightly lower precision of 94.1% compared to the rule-based classifier, but at a significantly better recall of 93.9% / Detta exjobb har undersökt problemet att detektera automatiserade Twitter-konton relaterade till Ukraina-konflikten genom att använda övervakade maskininlärningsmetoder. Ett slående problem med den insamlade datamängden var avsaknaden av träningsexempel. I övervakad maskininlärning brukar man traditionellt manuellt märka upp en träningsmängd. Detta medför dock långtråkigt arbete samt att det blir dyrt förstora och ständigt föränderliga datamängder. Vi presenterar en ny metod för att syntetiskt generera uppmärkt Twitter-data (klassifieringsetiketter) för detektering av automatiserade konton med en regel-baseradeklassificerare. Metoden medför en signifikant minskning av resurser och anstränging samt snabbar upp processen att anpassa klassificerare till förändringar i Twitter-domänen. En utvärdering av klassificerare utfördes på en manuellt uppmärkt testmängd bestående av 1,000 Twitter-konton. Resultaten visar att den regelbaserade klassificeraren på egen hand uppnår en precision på 94.6% och en recall på 52.9%. Vidare påvisar resultaten att klassificerare baserat på övervakad maskininlärning kunde lära sig från syntetiskt uppmärkt data. I bästa fall uppnår dessa maskininlärningsbaserade klassificerare en något lägre precision på 94.1%, jämfört med den regelbaserade klassificeraren, men med en betydligt bättre recall på 93.9%.

Работа посвящена изучению и анализу противостояния Саудовской Аравии и Ирана как основных претендентов на лидерство на Ближнем Востоке. Автор раскрывает специфику арабо-иранских отношений в контексте цивилизационного подхода, анализирует особенности борьбы за лидерство двух стран на фоне основных событий на Ближнем Востоке конца XX – начала XXI века и выявляет военно-политическую направленность саудовско-иранского противостояния в современных ближневосточных конфликтах – сирийском и йеменском. / The paper is dedicated to the study and analysis of the confrontation of Saudi Arabia and Iran as the main contenders for the leadership in the Middle East. The author reveals the character of the Arab-Iranian relations in the context of civilizational approach, analyzes the features of the two countries struggle for the leadership against the background of major developments in the Middle East in the period of the end of XX to the beginning of XXI centuries and reveals the political and military direction of Saudi-Iranian confrontation in the modern Middle East conflict such as the Syrian and Yemeni ones.