JSON-to-JSON transformation languages enable the transformation of a JSON document into another JSON document. As JSON is gradually becoming the most used interchange format on the Internet there is a need for transformation languages that can transform the data stored in JSON in order for the data to be used with other systems. The transformation can transform the document structurally, for example by altering the hierarchical structure of the document. The transformation can also transform the document textually, for example by renaming fields or altering values. None of the existing JSON-to-JSON transformation languages have become a standard (Jellife, 2017). This work evaluates the expressive power of the JSON-to-JSON transformation language Jolt. Jolt have recently been adopted by Apache and support have been introduced in some of their products. If a transformation language have expressive power that are at least equal to Nested Relational Algebra this implies that a transformation language can perform many advanced transformations. In this work  a formal model of Jolt is defined, referred to as Jolt0, in order to compare its expressive powers to Nested Relational Algebra. For that purpose, the operations of another formal model called MQuery which have been proven to have equivalent expressive power to Nested Relational Algebra are translated into Jolt0. It is shown that Jolt does not have expressive powers equivalent to Nested Relational Algebra. We further compared the performance of four JSON-to-JSON transformation languages (Jolt, Handlebars, Liquid, and XSLT 3.0) by constructing tests where the different transformation languages executed equivalent transformations. The transformations were evaluated by measuring runtime and memory usage. The study shows that XSLT 3.0 performed worst in all run time and memory usage tests. When transforming large input data XSLT 3.0 performed significantly worse than the other languages. / JSON-till-JSON transformationsspråk möjliggör transformationer från ett JSON-dokument till ett annat JSON-dokument. Eftersom JSON gradvis håller på att bli det mest använda data-utväxlingsformatet på internet så finns det ett behov av transformationsspråk som kan transformera data som är lagrad i JSON formatet för att kunna användas med andra system. Transformationen kan transformera dokumentet strukturellt, till exempel genom att förändra den hierarkiska strukturen på dokumentet. Transformationen kan även transformera dokumentet textuellt, till exempel genom att döpa om fält eller ändra värden. Ingen av de existerande JSON-till-JSON transformationsspråken har blivit en standard (Jellife, 2017). Det här arbetet undersöker uttryckskraften av Jolt vilket är ett JSON-till-JSON transformationsspråk. Jolt har nyligen fått stöd av Apache i några av deras produkter. Om ett transformationsspråk har en uttryckskraft som är ekvivalent med nästlad relationell algebra innebär det att språket kan utföra många avancerade transformationer. I det här arbetet definieras en formell modell av Jolt, kallad Jolt0, för att kunna jämföra dess uttryckskraft med nästlad relationell algebra. Till det syftet så översätts operationerna från en annan formell modell med namnet MQuery som har bevisats ha ekvivalent uttrykskraft med nästlad relationell algebra till Jolt0. Arbetet drar slutsatsen att Jolt inte har uttryckskraft som är ekvivalent med nästlad relationell algebra. Arbetet undersöker också prestandan för de fyra JSON-till-JSON transformationsspråken (Jolt, Handlebars, Liquid och XSLT 3.0) genom att konstruera tester där de olika transformationsspråken exekverar ekvivalenta transformationer. Transformationerna utvärderas baserat på körstids- och minnesanvändningsprestandan. Studien visar att XSLT 3.0 presterar sämst i alla körstids- och minnesanvändningstester. När transformationerna använder sig av stor input data så presterar XSLT 3.0 signifikant sämre än de andra språken.

In this thesis we investigate the expressive power of several logics over finite trees. In particular we want to understand precisely the expressive power of first-order logic over finite trees. Because we study many logics, we proceed by comparison to a logic that subsumes them all and serves as a yardstick: monadic second-order logic. Each logic we consider is a fragment of monadic second-order logic. MSO is linked to the theory of formal languages. To each logical formula corresponds a tree language, which is the language of trees satisfying this formula. Furthermore, given a logic we can associate a class of tree languages: the class of languages definable by a formula of this logic. In the setting of finite trees MSO corresponds exactly to the class of regular tree languages. Given a logic, we actually look for a decidable characterization of the class of languages defined in this logic. By decidable characterization, we mean an algorithm for solving the following problem: given as input a finite tree automaton, decide if the recognized language belongs to the class in question. We will actually obtain our decidable characterizations by exhibiting for each class a set of closure properties such that a language is in the class under investigation if and only if it satisfies these closure properties. Each such closure property is then shown to be decidable. Stating and proving such closure properties usually yields a solid understanding of the expressive power of the corresponding logic. The main open problem in this research area is to obtain a decidable characterization for the class of tree languages that are definable in first-order logic. We provide decidable characterizations for several fragments of FO. First we provide three decidable characterizations for classes of regular languages of trees of bounded rank. The first class we consider is the class of languages definable in the temporal logic EF+F^-1. It essentially navigates the trees using two modalities for moving to a descendant node or an ancestor node. The second class we consider is the class of trees of bounded rank definable using one quantifier alternation. The last class, is the class of languages definable using a boolean combination of existential first order formulas. In the setting of forests, we investigate the class of languages definable in first-order logic using only two variables and two prediactes corresponding respectively to the ancestor and following sibling relations. We provide a characterization for this logic. The last class for which we provide a decidable characterization is the class of locally testable language (LT). A language L is in LT if membership in L depends only on the presence or absence of neighborhoods of a certain fixed size in the tree. We define notions of LT for both unranked trees and trees of bounded rank by adapting the definition of neighborhood to each setting. Then we provide a decidable characterization for both notions of LT.

Dans cette thèse nous étudions le pouvoir d'expression de plusieurs logiques sur les arbres finis. En particulier, nous cherchons à obtenir une compréhension précise du pouvoir d'expression de la logique du premier ordre sur les arbres finis. Nous étudions un nombre important de logiques- pour cette raison nous procédons par comparaison avec une logique qui les contient et nous sert de référence: la logique monadique du second-ordre. Chaque logique que nous considérons est un fragment de la logique monadique du second ordre. MSO est liée à la théorie des langages formels. A chaque formule logique correspond un langage d'arbre: celui des arbres satisfaisant la formule. De plus, étant donné une logique nous pouvons lui associer une classe de langages d'arbres: la classe des langages définissables par une formule de cette logique. Dans le cadre des arbres finis, MSO correspond exactement à la classe des langages réguliers. Étant donné une logique, nous cherchons en fait à obtenir une caractérisation décidable de la classe de langages définissable par celle-ci. Par caractérisation décidable nous entendons un algorithme résolvant le problème suivant: pour un automate d'arbre finis, décider si le langage appartient à la classe en question. Nos caractérisations décidables sont en fait obtenue en exhibant pour chaque classe un ensemble de propriétés de clôture vérifiées par un langage si et seulement si celui-ci appartient à la classe en question. Nous montrons ensuite que chaque propriété de clôture est décidable. Énoncer et prouver de telles propriétés de clôture permet généralement d'obtenir une bonne compréhension du pouvoir de la logique correspondante. Le problème ouvert principal de ce domaine de recherche est l'obtention d'une caractérisation décidable pour la logique du premier ordre. Nous présentons des caractérisation décidables pour plusieurs fragment de FO. Nous commençons par la présentation de trois caractérisations décidable pour des classes de langages d'arbres de rang borné. La première classe que nous considérons est celle des langages définissables par la logique EF + F-1. Cette logique permet de naviguer dans l'arbre en se déplaçant soit vers un ancêtre, soit vers un descendant. La second classe est celle des arbres de rang borné définissables par la logique du premier ordre en n'utilisant qu'une seule alternance de quantificateurs. La dernière classe est celle des langages définissables par une combinaison booléenne de formules existentielles du premier ordre. Dans le cadre des forêts, nous étudions la classe des langages définissable par la logique du premier ordre à deux variables et deux prédicats correspondants respectivement à la relation ancêtre et la relation frère suivant. Nous présentons une caractérisation pour cette logique. La dernière classe pour laquelle nous présentons une caractérisation décidable est celle des langages localement testables (LT). UN langage est dans LT si l'appartenance d'un arbre à celui-ci ne dépends que des voisinages d'une certaine taille fixée dans l'arbre. / In this thesis we investigate the expressive power of several logics over finite trees. In particular we want to understand precisely the expressive power of first-order logic over finite trees. Because we study many logics, we proceed by comparison to a logic that subsumes them all and serves as a yardstick: monadic second-order logic. Each logic we consider is a fragment of monadic second-order logic. MSO is linked to the theory of formal languages. To each logical formula corresponds a tree language, which is the language of trees satisfying this formula. Furthermore, given a logic we can associate a class of tree languages: the class of languages definable by a formula of this logic. In the setting of finite trees MSO corresponds exactly to the class of regular tree languages. Given a logic, we actually look for a decidable characterization of the class of languages defined in this logic. By decidable characterization, we mean an algorithm for solving the following problem: given as input a finite tree automaton, decide if the recognized language belongs to the class in question. We will actually obtain our decidable characterizations by exhibiting for each class a set of closure properties such that a language is in the class under investigation if and only if it satisfies these closure properties. Each such closure property is then shown to be decidable. Stating and proving such closure properties usually yields a solid understanding of the expressive power of the corresponding logic. The main open problem in this research area is to obtain a decidable characterization for the class of tree languages that are definable in first-order logic. We provide decidable characterizations for several fragments of FO. First we provide three decidable characterizations for classes of regular languages of trees of bounded rank. The first class we consider is the class of languages definable in the temporal logic EF+F^-1. It essentially navigates the trees using two modalities for moving to a descendant node or an ancestor node. The second class we consider is the class of trees of bounded rank definable using one quantifier alternation. The last class, is the class of languages definable using a boolean combination of existential first order formulas. In the setting of forests, we investigate the class of languages definable in first-order logic using only two variables and two prediactes corresponding respectively to the ancestor and following sibling relations. We provide a characterization for this logic. The last class for which we provide a decidable characterization is the class of locally testable language (LT). A language L is in LT if membership in L depends only on the presence or absence of neighborhoods of a certain fixed size in the tree. We define notions of LT for both unranked trees and trees of bounded rank by adapting the definition of neighborhood to each setting. Then we provide a decidable characterization for both notions of LT.

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第22807号 / 情博第737号 / 新制||情||126(附属図書館) / 京都大学大学院情報学研究科システム科学専攻 / (主査)教授 下平 英寿, 教授 田中 利幸, 教授 鹿島 久嗣 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

This thesis is a detailed examination of the expressive power of valued constraints and related complexity questions. The valued constraint satisfaction problem (VCSP) is a generalisation of the constraint satisfaction problem which allows to describe a variety of combinatorial optimisation problems. Although most results are stated in this framework, they can be interpreted equivalently in the framework of, for instance, pseudo-Boolean polynomials, Gibbs energy minimisation, or Markov Random Fields. We take a result of Cohen, Cooper and Jeavons that characterises the expressive power of valued constraint in terms of certain algebraic properties, and extend this result by showing yet another connection between the expressive power of valued constraints and linear programming. We prove a decidability result for fractional clones. We consider various classes of valued constraints and the associated cost functions with respect to the question of which of these classes can be expressed using only cost functions of bounded arities. We identify the first known example of an infinite chain of classes of constraints with strictly increasing expressive power. We present a full classification of various classes of constraints with respect to this problem. We study submodular constraints and cost functions. Submodular functions play a key role in combinatorial optimisation and are often considered to be a discrete analogue of convex functions. It has previously been an open problem whether all Boolean submodular cost functions can be decomposed into a sum of binary submodular cost functions over a possibly larger set of variables. This problem has been considered within several different contexts in computer science, including computer vision, artificial intelligence, and pseudo-Boolean optimisation. Using a connection between the expressive power of valued constraints and certain algebraic properties of cost functions, we answer this question negatively. Our results have several corollaries. First, we characterise precisely which submodular polynomials of degree 4 can be expressed by quadratic submodular polynomials. Next, we identify a novel class of submodular functions of arbitrary arities that can be expressed by binary submodular functions, and therefore minimised efficiently using a so-called expressibility reduction to the (s,t)-Min-Cut problem. More importantly, our results imply limitations on this kind of reduction and establish for the first time that it cannot be used in general to minimise arbitrary submodular functions. Finally, we refute a conjecture of Promislow and Young on the structure of the extreme rays of the cone of Boolean submodular functions.

Recently, verification and analysis of data words have gained a lot of interest. Metric temporal logic (MTL) and timed propositional temporal logic (TPTL) are two extensions of Linear time temporal logic (LTL). In MTL, the temporal operator are indexed by a constraint interval. TPTL is a more powerful logic that is equipped with a freeze formalism. It uses register variables, which can be set to the current data value and later these register variables can be compared with the current data value. For monotonic data words, Alur and Henzinger proved that MTL and TPTL are equally expressive and the satisfiability problem is decidable. We study the expressive power, satisfiability problems and path checking problems for MLT and TPTL over all data words. We introduce Ehrenfeucht-Fraisse games for MTL and TPTL. Using the EF-game for MTL, we show that TPTL is strictly more expressive than MTL. Furthermore, we show that the MTL definability problem that whether a TPTL-formula is definable in MTL is not decidable. When restricting the number of register variables, we are able to show that TPTL with two register variables is strictly more expressive than TPTL with one register variable. For the satisfiability problem, we show that for MTL, the unary fragment of MTL and the pure fragment of MTL, SAT is not decidable. We prove the undecidability by reductions from the recurrent state problem and halting problem of two-counter machines. For the positive fragments of MTL and TPTL, we show that a positive formula is satisfiable if and only it is satisfied by a finite data word. Finitary SAT and infinitary SAT coincide for positive MTL and positive TPTL. Both of them are r.e.-complete. For existential TPTL and existential MTL, we show that SAT is NP-complete. We also investigate the complexity of path checking problems for TPTL and MTL over data words. These data words can be either finite or infinite periodic. For periodic words without data values, the complexity of LTL model checking belongs to the class AC^1(LogDCFL). For finite monotonic data words, the same complexity bound has been shown for MTL by Bundala and Ouaknine. We show that path checking for TPTL is PSPACE-complete, and for MTL is P-complete. If the number of register variables allowed is restricted, we obtain path checking for TPTL with only one register variable is P-complete over both infinite and finite data words; for TPTL with two register variables is PSPACE-complete over infinite data words. If the encoding of constraint numbers of the input TPTL-formula is in unary notation, we show that path checking for TPTL with a constant number of variables is P-complete over infinite unary encoded data words. Since the infinite data word produced by a deterministic one-counter machine is periodic, we can transfer all complexity results for the infinite periodic case to model checking over deterministic one-counter machines.

FERREIRA, Francicleber Martins. Modelos minimais e hierarquia de expressividade. 2007. 109 f. : Dissertação (mestrado) - Universidade Federal do Ceará, Centro de Ciências, Departamento de Computação, Fortaleza-CE, 2007. / Submitted by guaracy araujo (guaraa3355@gmail.com) on 2016-06-29T18:55:46Z No. of bitstreams: 1 2007_dis_fmferreira.pdf: 752533 bytes, checksum: 98c57917ae2bd6af5de38f25d9ce7c39 (MD5) / Approved for entry into archive by guaracy araujo (guaraa3355@gmail.com) on 2016-06-29T18:56:33Z (GMT) No. of bitstreams: 1 2007_dis_fmferreira.pdf: 752533 bytes, checksum: 98c57917ae2bd6af5de38f25d9ce7c39 (MD5) / Made available in DSpace on 2016-06-29T18:56:33Z (GMT). No. of bitstreams: 1 2007_dis_fmferreira.pdf: 752533 bytes, checksum: 98c57917ae2bd6af5de38f25d9ce7c39 (MD5) Previous issue date: 2007 / Neste trabalho, o conceito de Modelo Minimal e seu uso na semântica de certas lógicas são estudados. Nós analisamos o poder expressivo de diversas lógicas que usam o conceito de Modelo Minimal para definir sua relação de satisfação. Os principais teoremas estudados foram o Teorema de Löwenheim-Skolem e o Teorema de Definibilidade de Beth. No Capítulo 1, nós damos algumas motivações e revisamos alguns conceitos básicos de Lógica. No Capítulo 2, nos estudamos a Lógica de Menor Ponto Fixo|LFP. Nós exibimos uma prova de que o Teorema de Beth não vale para LFP. Nós usamos teorias infinitas para provar isso. Utilizando um resultado de Hodkinson para L!!1!, nós mostramos que o Teorema de Beth continua não valendo mesmo para teorias finitas de LFP. Nós continuamos estudando problemas de definibilidade para LFP e demonstramos que, para tipos especiais de definições implícitas formadas por Sistemas Recursivos, que funcionam como definições recursivas em determinados contextos, existe uma definição explícita. Nós promavos ainda que o Teorema de LÄowenheim-Skolem Descendente vale para qualquer conjunto de fórmulas de LFP, independentemente de sua cardinalidade. No Capítulo 3, a Circunscrição de McCarthy e as Teorias Circunscritivas Aninhadas de Lifschitz, uma generalização da primeira. Nós abordamos o poder expressivo de Circunscrição e a falha do Teorema de LÄowenheim-Skolem Descendente. Nós também investigamos questões de definibilidade no contexto de Circunscrição. Nós encerramos esse capítulo mostrando que as Teorias Circunscritivas Aninhadas possuem poder expressivo comparável com o da Lógica de Segunda-Ordem. No Capítulo 4, nós estendemos uma lógica criada por van Benthem dando origem a duas outras lógicas, a saber, U-MIN e I-MIN. Nós provamos que ambas são equivalentes entre si em poder expressivo e daí em diante chamamos U-MIN de MIN. Nós introduzimos a Lógica Si-MIN de minimalização simultânea e provamos que Si-MIN é equivalente a U-MIN e I-MIN e também à Lógica de Segunda-Ordem. Nós então propomos o fragmento MIN¢ de MIN, cujo poder expressivo situa-se entre o da Lógica de Segunda-Ordem e o de LFP. No Capítulo 5, nós reunimos nossas conclusões e apontamos trabalhos futuros.

Conselho Nacional de Desenvolvimento CientÃfico e TecnolÃgico / Neste trabalho, o conceito de Modelo Minimal e seu uso na semÃntica de certas lÃgicas sÃo estudados. NÃs analisamos o poder expressivo de diversas lÃgicas que usam o conceito de Modelo Minimal para definir sua relaÃÃo de satisfaÃÃo. Os principais teoremas estudados foram o Teorema de LÃwenheim-Skolem e o Teorema de Definibilidade de Beth. No CapÃtulo 1, nÃs damos algumas motivaÃÃes e revisamos alguns conceitos bÃsicos de LÃgica. No CapÃtulo 2, nos estudamos a LÃgica de Menor Ponto Fixo|LFP. NÃs exibimos uma prova de que o Teorema de Beth nÃo vale para LFP. NÃs usamos teorias infinitas para provar isso. Utilizando um resultado de Hodkinson para L!!1!, nÃs mostramos que o Teorema de Beth continua nÃo valendo mesmo para teorias finitas de LFP. NÃs continuamos estudando problemas de definibilidade para LFP e demonstramos que, para tipos especiais de definiÃÃes implÃcitas formadas por Sistemas Recursivos, que funcionam como definiÃÃes recursivas em determinados contextos, existe uma definiÃÃo explÃcita. NÃs promavos ainda que o Teorema de LÃowenheim-Skolem Descendente vale para qualquer conjunto de fÃrmulas de LFP, independentemente de sua cardinalidade. No CapÃtulo 3, a CircunscriÃÃo de McCarthy e as Teorias Circunscritivas Aninhadas de Lifschitz, uma generalizaÃÃo da primeira. NÃs abordamos o poder expressivo de CircunscriÃÃo e a falha do Teorema de LÃowenheim-Skolem Descendente. NÃs tambÃm investigamos questÃes de definibilidade no contexto de CircunscriÃÃo. NÃs encerramos esse capÃtulo mostrando que as Teorias Circunscritivas Aninhadas possuem poder expressivo comparÃvel com o da LÃgica de Segunda-Ordem. No CapÃtulo 4, nÃs estendemos uma lÃgica criada por van Benthem dando origem a duas outras lÃgicas, a saber, U-MIN e I-MIN. NÃs provamos que ambas sÃo equivalentes entre si em poder expressivo e daà em diante chamamos U-MIN de MIN. NÃs introduzimos a LÃgica Si-MIN de minimalizaÃÃo simultÃnea e provamos que Si-MIN à equivalente a U-MIN e I-MIN e tambÃm à LÃgica de Segunda-Ordem. NÃs entÃo propomos o fragmento MIN de MIN, cujo poder expressivo situa-se entre o da LÃgica de Segunda-Ordem e o de LFP. No CapÃtulo 5, nÃs reunimos nossas conclusÃes e apontamos trabalhos futuros.

Recently, verification and analysis of data words have gained a lot of interest. Metric temporal logic (MTL) and timed propositional temporal logic (TPTL) are two extensions of Linear time temporal logic (LTL). In MTL, the temporal operator are indexed by a constraint interval. TPTL is a more powerful logic that is equipped with a freeze formalism. It uses register variables, which can be set to the current data value and later these register variables can be compared with the current data value. For monotonic data words, Alur and Henzinger proved that MTL and TPTL are equally expressive and the satisfiability problem is decidable. We study the expressive power, satisfiability problems and path checking problems for MLT and TPTL over all data words. We introduce Ehrenfeucht-Fraisse games for MTL and TPTL. Using the EF-game for MTL, we show that TPTL is strictly more expressive than MTL. Furthermore, we show that the MTL definability problem that whether a TPTL-formula is definable in MTL is not decidable. When restricting the number of register variables, we are able to show that TPTL with two register variables is strictly more expressive than TPTL with one register variable. For the satisfiability problem, we show that for MTL, the unary fragment of MTL and the pure fragment of MTL, SAT is not decidable. We prove the undecidability by reductions from the recurrent state problem and halting problem of two-counter machines. For the positive fragments of MTL and TPTL, we show that a positive formula is satisfiable if and only it is satisfied by a finite data word. Finitary SAT and infinitary SAT coincide for positive MTL and positive TPTL. Both of them are r.e.-complete. For existential TPTL and existential MTL, we show that SAT is NP-complete. We also investigate the complexity of path checking problems for TPTL and MTL over data words. These data words can be either finite or infinite periodic. For periodic words without data values, the complexity of LTL model checking belongs to the class AC^1(LogDCFL). For finite monotonic data words, the same complexity bound has been shown for MTL by Bundala and Ouaknine. We show that path checking for TPTL is PSPACE-complete, and for MTL is P-complete. If the number of register variables allowed is restricted, we obtain path checking for TPTL with only one register variable is P-complete over both infinite and finite data words; for TPTL with two register variables is PSPACE-complete over infinite data words. If the encoding of constraint numbers of the input TPTL-formula is in unary notation, we show that path checking for TPTL with a constant number of variables is P-complete over infinite unary encoded data words. Since the infinite data word produced by a deterministic one-counter machine is periodic, we can transfer all complexity results for the infinite periodic case to model checking over deterministic one-counter machines.

Dans cette thèse, nous avons étudié d’une manière structurée et détaillée la sémantique des verbes gnat’ et gonjat’. Pour cela, nous avons tout d’abord formalisé leur potentiel sémantique qui comprend les composants de leur valeur de base et ce qu’ils génèrent dans des situations nouvelles, dans des cotextes particuliers. Pour chaque valeur sémantique, nous avons présenté un schéma sémantique, un schéma actantiel en précisant les rôles sémantiques, la nature des actants et leur réalisation syntaxique, ainsi que des illustrations d’emploi dans des cotextes des plus représentatifs, des synonymes, la famille morphologique. Les valeurs plus ou moins communes à gnat’ et à gonjat’ ont été étudiées ensemble, ce qui nous a permis de mettre en évidence leurs similitudes et différences. Puis, nous avons traité la question de la catégorie grammaticale des verbes de mouvement. Après avoir exposé les propriétés traditionnellement attribuées aux corrélats déterminé / indéterminé et les particularités qui sont générées dans l’emploi de chacun de ces corrélats, nous avons analysé la corrélation déterminé / indéterminé en général, et en particulier, entre gnat’ et gonjat’. Faisant partie de la catégorie grammaticale des verbes de mouvement, gnat’ et gonjat’ s’opposent sur le plan grammatical en tant que déterminé vs indéterminé et leurs caractéristiques grammaticales induisent quelques différences sémantiques. Ainsi, la plupart des différences observées entre gnat’ et gonjat’ ne sont pas en fait pas d’ordre lexical, mais résultent des différences grammaticales du déterminé gnat’ et de l’indéterminé gonjat’. Dans la partie consacrée à la dérivation morphologique, nous avons eu pour objectif d’examiner les effets de la préverbation et les modifications de sens que la préverbation entraîne pour ces verbes. Nous avons étudié les préverbés formés sur la base de gnat’ / gonjat’ à valeur spatiale en les regroupant en fonction de l’orientation spatiale : à partir du point de départ, vers le point d’arrivée, les deux en même temps ou en rapport avec un repère. Une attention minutieuse a été accordée aux différences sémantiques et aspectuelles de certains préverbés partageant les mêmes valeurs spatiales. L’étude des préverbés combinés aux préverbes à valeur non spatiale a été organisée en fonction de la morphologie de l’aspect : les préverbés hors couple et les couples de verbes. Quelques particularités ont été observées. Pour que l’étude sémantique soit complète, nous y avons inclus la question des phrasèmes comportant ces verbes. Puis, nous avons analysé le parallélisme qui existe entre gnat’ / gonjat’ simples et préverbés et le verbe dit de position sidet’ : les valeurs que ces verbes ont en commun et aussi, les limites de leur corrélation, Enfin, en partant directement des situations linguistiques concrètes avec gnat’ / gonjat’ simples et préverbés nous avons exposé un certain nombre de mécanismes qui aboutissent à un effet expressif. / In this thesis, we have studied the semantics of the verbs gnat’ and gonjat’ in a structured and extensive way. To achieve this, we began by structuring their semantic potential, which includes their basic value and what it produces in new situations, in specific cotexts. For each semantic value, we have presented a semantic schema, mapping the semantic roles of the actants, their nature and their syntactic development, together with examples of usage in the most representative cotexts, the synonyms, the morphological family. The more or less common values of gnat’ and gonjat’ have been studied together, so that we could bring their similarities and differences to light. Then, we dealt with the question of the grammatical category of the verbs of motion. Having presented the usually attributed properties of the correlate determined / indetermined and the particularities that emerge in the usage of each of these correlates, we have focused on the determined / indetermined correlation in general and between gnat’ and gonjat’ in particular. While belonging to the same grammatical category of the verbs of motion, both gnat’ and gonjat’ are opposed on a grammatical level, as determined and indetermined, and their grammatical characteristics lead to some semantical differences. Hence, most of the observed differences between gnat’ and gonjat’ are not of a lexical type; they are rather the result of the semantical differences between the determined gnat’ and the indetermined gonjat’. In the part of our thesis dedicated to morphological derivation, our goal was to study the impacts of verbal prefixation and the change of meaning as a result of the prefixation for these verbs. We have studied prefixed verbs based on gnat’ / gonjat’ with spatial meanings, clustering them depending on their spatial orientation: from the starting point, to the point of arrival, both at the same time or in relation to a landmark. A particular focus has been placed on the semantical and aspectual differences of some verbs with prefixes that share the same spatial values. The analysis of verbs with spatial prefixes was done according to the morphology of the aspect: the verbs with prefixes outside aspectual pairs and the pair of verbs. Some particularities have been observed. For the semantic study to be complete, we have added the question of the phrasemes using these verbs. Then, we studied the similarity between gnat’ and gonjat’ in their simple and prefixed forms and the verb of position sidet’: what those verbs have in common but also the limits of their similarities. Finally, we have taken concrete linguistic situations involving gnat’ and gonjat’ in their simple and prefixed forms in order to outline a number of mechanisms leading to an expressive result.