過去由於我國立法委員選舉所採用的「複數選區單記非讓渡投票制」，常造成選風敗壞、偏激取向、派系政治等諸多負面影響。有鑑於此，立法院於2004年8月23日通過「席次減半」及「單一選區兩票制」的修憲案，並於2005年6月7日經國民大會複決通過，自第七屆開始，立法委員任期改為4年，席次減半為113人。2008年1月12日第七屆立法委員選舉首次實施「單一選區兩票制」，而本文將檢視此次立委選舉結果相關資料，探討選制改變後對我國政黨體系之影響，並從過去的理論評析與實際運作層面相互驗證，探討理論與實務之間異同，及其可能之影響因素，然而，由於僅是一次實施的結果，因此相關後續發展仍須持續觀察。 / In the past, our national elections of members of Legislative Yuan adopted “multi-member-district, single non-transferable vote.” It usually caused negative influences such as deterioration of election, extreme trends, and faction politics. Therefore, Legislative Yuan passed constitution-amending bills “seats reduction in half” and “single-district two votes system” on August 23rd, 2004. These bills were passed by National Assembly on June 7th, 2005. The term of service for the seventh legislators was prolonged into four years and the seats were reduced in half to 113 seats. The seventh legislator election adopted “single-district two votes system” for the first time on January 12th, 2008. The essay will survey information related to the outcome of the legislator election, discussing the influence of changing election systems on the systems of political parties. Through the mutual demonstration between past theory criticism and practical operation, it discusses common and different points between theories and practices and finds out possible influential elements. However, it is just the outcome of single practice so it’s following development requires constant observations.

L’objet de ce travail est d’étudier les mécanismes de co-variation entre la tête d’un constituant nominal et son/ses dépendant(s) dans les langues indo-européennes anciennes : indo-iranien, grec, italique et anatolien. Une analyse détaillée de la morphologie nominale et pronominale au sein de chacun de ces groupes permet de dégager les classes d’accord possibles et impossibles pour chacune de ces langues et de déterminer le rôle joué par les morphèmes désinentiels dans la co-variation. Dans les langues à trois genres, les classes flexionnelles sont plus nombreuses avec une information sur le genre portée tantôt par la tête, tantôt par le dépendant, ou parfois demeure ambiguë. En revanche, lorsque le dépendant est un pronom, le type morphologique en général implique le genre.Une attention particulière est accordée aux relations d’accord entre tête et dépendant dans les rôles syntaxiques élémentaires : A (agent transitif), P (patient), U (participant unique d’une construction intransitive). Le marquage de ces rôles diffère notablement entre les noms de genre animé et ceux de genre inanimé. Le marquage des animés présente les mêmes caractéristiques dans les quatre groupes, tandis que pour les inanimés on observe que l’anatolien a un comportement radicalement distinct : le marquage différencié de A et de U montre une capacité limitée des noms inanimés à accéder au rôle d’agent, également reflétée dans l’indexation au verbe.On estime que l’organisation dont l’anatolien témoigne doit être posée comme originelle dans le domaine indo-européen : la contrainte de l’animation est déterminante pour accéder au rôle d’agent. / The purpose of this work is to study co-variation mechanisms between the head of a nominal constituent and its dependent(s) in the following ancient Indo-European languages: Indo-Iranian, Greek, Italic and Anatolian. By analysing in detail nominal and pronominal morphology within each of these groups, we shall determine which agreement classes are possible, and which ones are not, and define which role is played by desinential morphemes in the co-variation. In languages which distinguish three genders, inflectional classes are more numerous and either the head or the dependent bear the gender information, or neither, or both. However, when the dependent is a pronoun, the gender may be implied by the morphological type.We shall particularly focus on the agreement relations between head and dependent in core syntactic roles: A (transitive agent), P (patient) and U (unique participant in an intransitive construction). The marking of these roles is considerably different depending on whether the nouns are animate or inanimate. Animate nouns are marked in the same way in the four groups which are under scrutiny, whereas in the case of inanimate nouns Anatolian behaves in a drastically different way: the differential marking of A and U shows a limited capacity of inanimate nouns to become agents, which is also reflected in verbal indexation.The organisation shown by Anatolian is postulated as original in the field of Indo-European: the constraint of animation is decisive in becoming an agent.

Cette recherche traite de la réactivité des parlementaires français vis-à-vis de leur circonscription.Ce faisant, notre thèse investit l’aspect dyadique de la représentation politique qui a été largementignoré par la littérature française et offre ainsi un nouveau regard sur la variété des activités desdéputés français tant à Paris qu’en circonscription. En s’inspirant des écrits du néoinstitutionnalismedes choix rationnels, notre thèse vise à mettre en lumière les déterminantsinstitutionnels de la réactivité vis-à-vis de la circonscription ainsi que les logiques qui sous-tendentles pratiques de représentation. En surmontant le descriptivisme de beaucoup d’études à ce sujetet à l’aide de solides matériaux empiriques, nous montrons que le territoire d’élection affectegrandement le comportement des députés tant dans le contenu que sur la forme des activités. Lacirconscription et les attentes différenciées des citoyens agissent de manière systématique sur lamanière dont les députés mènent leur mandat. / This research deals with French MPs responsiveness. This dissertation studies the dyadicrepresentation in the French Assemblée nationale which was largely neglected by scholars. It givesthus a new insight on the diversity of French MPs activities both in Paris and in their district.Following the arguments of the rational choice new institutionalism, our dissertation aims tohighlight the institutional determinants of MPs responsiveness toward their district and the logicsthat underlie their practices of representations. By overcoming the descriptiveness of many studiesand thanks to original and solid empirical data, we show that the characteristics of the districtstrongly affect both the kind and the content of MPs activities. The district and the various citizens’expectations systematically affect the way French MPs conceive their mandate.

This thesis investigates the formation of power structures in a revolutionary setting. It takes as a case study the central Siberian city of Krasnoiarsk, in which a powerful Soviet of Workers’ and Soldiers’ Deputies emerged during the period March-October 1917. The Krasnoiarsk Soviet was an elective council established during the overthrow of Tsarist authorities. Throughout 1917, it became a vital component of an emerging local and regional power structure, assuming growing responsibility for a number of core state tasks. As well as providing a new empirical case study to English-language literature on 1917, the thesis employs a nuanced analytical approach which challenges existing conceptualisations of state power in revolution and the role played by local soviets. State power in revolutionary Russia has often been viewed as something to be contested between different political groupings and organisations seeking to assert their own outright control. This view is captured neatly by the formulation of “dual power”, in which soviets and Provisional Government organisations constructed alternative power bases in an attempt to wield outright control. Accordingly, the soviets’ growing political strength indicated an ability to marginalise other groups and organisations seeking to wield power. By contrast, this thesis does not seek to explain how power in revolutionary Krasnoiarsk was “captured” or otherwise controlled by the Soviet alone. Instead, it applies a critical interpretation of state power proposed by Bob Jessop and other theorists, who view the state as a site of interaction and negotiation between multiple autonomous organisations and social actors, all of which have a stake in the way it operates in practice. It focuses on the emergence of a “soviet power” writ small, in which the Krasnoiarsk Soviet became an authoritative organisation within a broader constellation of revolutionary actors. Without denying the Soviet’s centrality within this power structure, the thesis does not explain its role simply as the monopolisation of authority over other would-be contenders. Rather, it sees the Soviet’s importance in its ability to establish itself as a focal point for interactions between multiple actors which, collectively, shaped state power at a local and regional level. It considers how the forms and practices of revolutionary power developed through these interactions and how these interactions in turn transformed the roles of actors and organisations engaging them. In order to unpick the complex and dynamic processes of revolutionary power, the thesis employs three core methodological concepts: institutions, mobilisation, and ideology. It makes several important and original arguments. Firstly, it emphasises the autonomy of social actors which supported the Soviet and engaged in its politics, demonstrating the extent to which they were able to shape its political functions and structures according to their own concerns. Secondly, it reveals the importance of skilled administrative personnel to Soviet work, highlighting the invaluable practical roles they played in the regulation of provisions and their ability to influence Soviet policy measures on this issue. Thirdly, it demonstrates the close cooperation between the Soviet and other local governmental and administrative bodies, including the city Duma and provisions regulatory organisations, which remained vital to fulfilling state functions throughout 1917. Finally, it discusses how the Soviet and socialist activists challenged established power relationships between Krasnoiarsk, as a locality, and all-Russian state authorities, revealing the growing importance they attached to securing greater local autonomy in revolution and the changing ways local actors viewed their role in wider all-Russian politics.

I den här masteruppsatsen frågar vi oss vilka faktorer som har betydelse för riksdagsmandatets representation. En stor del av den allmänna bilden är att svensk politik i huvudsak är nationellt partiorienterad och att basen för folkviljans förverkligande är åsiktsrepresentation. Tidigare forskning tycks också ge stöd åt detta antagande. Uppsatsen frågar sig om detta verkligen är fallet. I studien används en kvalitativ intervjumetod för att ta sig ann forskningsproblemet. För att närmare förstå frågan använder vi oss också av en bredd av respondenter som inkluderar inte enbart riksdagsledamöter utan även regionala partiföreträdare. Förhoppningen är att detta ska ge en tydligare bild av hur olika politiska företrädare förstår riksdagsuppdragets representation.Med hjälp av representationsteori som referens visar analysen att även den regionala nivån har betydelse för riksdagsuppdraget. Både riksdagsledamöter och regionala partiföreträdare öppnar upp för att, även om den nationella nivåns betydelse inte är oviktig behöver man också införliva den regionala nivån och erkänna den betydelse. Detta tar bland annat uttryck i att så länge den nationella partilinjen fortfarande formas finns en förväntan att riksdagsledamöter ska driva länsfrågor och detta är något som ledamöter även menar sig göra. Men när den gemensamma partilinjen har formats finns också en utbredd acceptans för att riksdagsledamöter måste hålla sig till denna.Här finns en grund för fortsatt diskussion och forskning om vad som är fundamentet för den parlamentariska representationen i Sverige. Vad folkviljan baseras på och hur denna representation förstås är en hörnsten i hur vi förstår svensk politik. Genom att öppna upp för att mer än en enskild faktor är av betydelse då man definierar riksdagsuppdraget är uppsatsens viktigaste bidrag följande; att enbart förstå och tolka riksdagsuppdraget in i den nationella partipolitikens ramar blir en alltför snäv förståelse. Istället behöver man väga in fler faktorer för att förstå hur folkviljan förverkligas och som denna uppsats visar är den regionala nivån ett viktigt exempel på detta. / In this master's thesis, we ask which factors are important for the understanding of the representation of the parliamentary mandate in Sweden. A large part of the general picture is that Swedish politics is mainly nationally party-oriented and that the basis for the realization of the will of the people is representation of opinion. Previous research also seems to support this assumption. The essay asks if this is really the case. The study uses a qualitative interview method to address the research problem. To better understand the issue, we also use a breadth of respondents that includes not only members of parliament (MPs) but also regional party representatives. The hope is that this will provide a clearer picture of how different political representatives understand the parliamentary representation.With the help of representation theory as a reference, the analysis shows that the regional level is also important for the representation of MPs. Both members of the Riksdag and regional party representatives are open to the fact that, although the importance of the national level is not unimportant, it is also necessary to incorporate the regional level and recognize its importance. This is expressed, among other things, in the fact that as long as the national party line is still being formed, there is an expectation that members of the Riksdag will pursue county issues, and this is something that MPs also claim’s to do. But when the central party line has been formed, there is also widespread acceptance that MPs must adhere to it.Here is a basis for further discussion and research on what is the foundation for parliamentary representation in Sweden. What the will of the people is based on and how this representation is understood is a cornerstone in how we understand Swedish politics. By opening to the fact that more than one individual factor is important when defining political representation, the thesis' most important contribution is the following; to only understand and interpret representation within the framework of national party politics becomes an overly narrow understanding. Instead, more factors need to be considered to understand how the will of the people is realized and as this essay shows, the regional level is an important example of this.

Cette thèse par article comprend quatre articles qui contribuent au domaine de l'apprentissage profond, en particulier à l'accélération de l’apprentissage par le biais de réseaux à faible précision et à l'application de réseaux de neurones profonds au traitement du langage naturel. Dans le premier article, nous étudions un schéma d’entraînement de réseau de neurones qui élimine la plupart des multiplications en virgule flottante. Cette approche consiste à binariser ou à ternariser les poids dans la propagation en avant et à quantifier les états cachés dans la propagation arrière, ce qui convertit les multiplications en changements de signe et en décalages binaires. Les résultats expérimentaux sur des jeux de données de petite à moyenne taille montrent que cette approche produit des performances encore meilleures que l’approche standard de descente de gradient stochastique, ouvrant la voie à un entraînement des réseaux de neurones rapide et efficace au niveau du matériel. Dans le deuxième article, nous avons proposé un mécanisme structuré d’auto-attention d’enchâssement de phrases qui extrait des représentations interprétables de phrases sous forme matricielle. Nous démontrons des améliorations dans 3 tâches différentes: le profilage de l'auteur, la classification des sentiments et l'implication textuelle. Les résultats expérimentaux montrent que notre modèle génère un gain en performance significatif par rapport aux autres méthodes d’enchâssement de phrases dans les 3 tâches. Dans le troisième article, nous proposons un modèle hiérarchique avec graphe de calcul dynamique, pour les données séquentielles, qui apprend à construire un arbre lors de la lecture de la séquence. Le modèle apprend à créer des connexions de saut adaptatives, ce qui facilitent l'apprentissage des dépendances à long terme en construisant des cellules récurrentes de manière récursive. L’entraînement du réseau peut être fait soit par entraînement supervisée en donnant des structures d’arbres dorés, soit par apprentissage par renforcement. Nous proposons des expériences préliminaires dans 3 tâches différentes: une nouvelle tâche d'évaluation de l'expression mathématique (MEE), une tâche bien connue de la logique propositionnelle et des tâches de modélisation du langage. Les résultats expérimentaux montrent le potentiel de l'approche proposée. Dans le quatrième article, nous proposons une nouvelle méthode d’analyse par circonscription utilisant les réseaux de neurones. Le modèle prédit la structure de l'arbre d'analyse en prédisant un scalaire à valeur réelle, soit la distance syntaxique, pour chaque position de division dans la phrase d'entrée. L'ordre des valeurs relatives de ces distances syntaxiques détermine ensuite la structure de l'arbre d'analyse en spécifiant l'ordre dans lequel les points de division seront sélectionnés, en partitionnant l'entrée de manière récursive et descendante. L’approche proposée obtient une performance compétitive sur le jeu de données Penn Treebank et réalise l’état de l’art sur le jeu de données Chinese Treebank. / This thesis by article consists of four articles which contribute to the field of deep learning, specifically in the acceleration of training through low-precision networks, and the application of deep neural networks on natural language processing. In the first article, we investigate a neural network training scheme that eliminates most of the floating-point multiplications. This approach consists of binarizing or ternarizing the weights in the forward propagation and quantizing the hidden states in the backward propagation, which converts multiplications to sign changes and binary shifts. Experimental results on datasets from small to medium size show that this approach result in even better performance than standard stochastic gradient descent training, paving the way to fast, hardware-friendly training of neural networks. In the second article, we proposed a structured self-attentive sentence embedding that extracts interpretable sentence representations in matrix form. We demonstrate improvements on 3 different tasks: author profiling, sentiment classification and textual entailment. Experimental results show that our model yields a significant performance gain compared to other sentence embedding methods in all of the 3 tasks. In the third article, we propose a hierarchical model with dynamical computation graph for sequential data that learns to construct a tree while reading the sequence. The model learns to create adaptive skip-connections that ease the learning of long-term dependencies through constructing recurrent cells in a recursive manner. The training of the network can either be supervised training by giving golden tree structures, or through reinforcement learning. We provide preliminary experiments in 3 different tasks: a novel Math Expression Evaluation (MEE) task, a well-known propositional logic task, and language modelling tasks. Experimental results show the potential of the proposed approach. In the fourth article, we propose a novel constituency parsing method with neural networks. The model predicts the parse tree structure by predicting a real valued scalar, named syntactic distance, for each split position in the input sentence. The order of the relative values of these syntactic distances then determine the parse tree structure by specifying the order in which the split points will be selected, recursively partitioning the input, in a top-down fashion. Our proposed approach was demonstrated with competitive performance on Penn Treebank dataset, and the state-of-the-art performance on Chinese Treebank dataset.