Cette thèse porte sur la composition des services de données et l'étude de l'impact de l'incertitude qui peut être associée à leurs données accessibles sur le processus de composition et de sélection de service. En effet, dans un contexte tel que l'Internet, il est de plus en plus reconnu que les données et les services d'accès aux données sont sujettes à des valeurs d'incertitude tout en exigeant des techniques de gestion plus sophistiquées. Dans cette thèse, nous enrichissons la description sémantique des services Web afin de refléter l'incertitude, et nous proposons de nouveaux mécanismes et modèles pour la sélection et la composition des services. Nos mécanismes sont basés sur les ensembles flous et les théories probabilistes. Tout d'abord, nous étendons notre modélisation précédente basée sur les vues RDF afin d'inclure les contraintes floues qui caractérisent les données accédées par les services. Nous proposons une algèbre de composition qui permet de classer les résultats retournés en fonction de leur pertinence par rapport aux préférences de l'utilisateur. Notre algèbre proposée repose sur les fondations de bases de données floues. En outre, nous optons pour l'approche probabiliste pour modéliser l'incertitude des données renvoyées par les services incertains. Nous étendons la description du service Web standard pour représenter les probabilités de sortie. L'invocation des services est également étendue pour tenir compte de l'incertitude. Cette extension est basée sur la théorie des mondes possibles utilisée dans les bases de données probabiliste. Nous définissons un ensemble d'opérateurs de composition qui sont nécessaires pour orchestrer les services de données. Pour chaque composition, plusieurs plans d'orchestration peuvent être possibles mais qui sont pas tous corrects, donc, nous définissons un ensemble de conditions pour vérifier si le plan est correct (Safe) ou pas. Nous fournissons une implémentation de nos différentes techniques et les appliquer aux domaines de l'immobilier et du commerce électronique. Nous implémentons ces services et nous fournissons également une étude de la performance de notre prototype de composition / In this thesis we focus on the data web services composition problem and study the impact of the uncertainty that may be associated with the output of a service on the service selection and composition processes. This work is motivated by the increasing number of application domains where data web services may return uncertain data, including the e-commerce, scientific data exploration, open web data, etc. We call such services that return uncertain data as uncertain services. In this dissertation, we propose new models and techniques for the selection and the composition of uncertain data web services. Our techniques are based on well established fuzzy and probabilistic database theories and can handle the uncertainty efficiently. First, we proposed a composition model that takes into account the user preferences. In our model, user preferences are modelled as fuzzy constraints, and services are described with fuzzy constraints to better characterize their accessed data. The composition model features also a composition algebra that allows us to rank the returned results based on their relevance to user's preferences. Second, we proposed a probabilistic approach to model the uncertainty of the data returned by uncertain data services. Specifically, we extended the web service description standards (e.g., WSDL) to represent the outputs' probabilities. We also extended the service invocation process to take into account the uncertainty of input data. This extension is based on the possible worlds theory used in the probabilistic databases. We proposed also a set of probability-aware composition operators that are necessary to orchestrate uncertain data services. Since a composition may accept multiple orchestration plans and not all of them compute the correct probabilities of outputs, we defined a set of conditions to check if a plan is safe (i.e., computes the probabilities correctly) or not. We implemented our different techniques and applied them to the real-estate and e-commerce domains. We provide a performance study of our different composition techniques

Nous proposons une approche pour ajouter des propriétés non-fonctionnelles (traitement d'exceptions, atomicité, sécurité, persistance) à des coordinations de services. L'approche est basée sur un Modèle de Politiques Actives (AP Model) pour représenter les coordinations de services avec des propriétés non-fonctionnelles comme une collection de types. Dans notre modèle, une coordination de services est représentée comme un workflow compose d'un ensemble ordonné d'activité. Chaque activité est en charge d'implante un appel à l'opération d'un service. Nous utilisons le type Activité pour représenter le workflow et ses composants (c-à-d, les activités du workflow et l'ordre entre eux). Une propriété non-fonctionnelle est représentée comme un ou plusieurs types de politiques actives, chaque politique est compose d'un ensemble de règles événement-condition-action qui implantent un aspect d'un propriété. Les instances des entités du modèle, politique active et activité peuvent être exécutées. Nous utilisons le type unité d'exécution pour les représenter comme des entités dont l'exécution passe par des différents états d'exécution en exécution. Lorsqu'une politique active est associée à une ou plusieurs unités d'exécution, les règles vérifient si l'unité d'exécution respecte la propriété non-fonctionnelle implantée en évaluant leurs conditions sur leurs états d'exécution. Lorsqu'une propriété n'est pas vérifiée, les règles exécutant leurs actions pour renforcer les propriétés en cours d'exécution. Nous avons aussi proposé un Moteur d'exécution de politiques actives pour exécuter un workflow orientés politiques actives modélisé en utilisant notre AP Model. Le moteur implante un modèle d'exécution qui détermine comment les instances d'une AP, une règle et une activité interagissent entre elles pour ajouter des propriétés non-fonctionnelles (NFP) à un workflow en cours d'exécution. Nous avons validé le modèle AP et le moteur d'exécution de politiques actives en définissant des types de politiques actives pour adresser le traitement d'exceptions, l'atomicité, le traitement d'état, la persistance et l'authentification. Ces types de politiques actives ont été utilisés pour implanter des applications à base de services fiables, et pour intégrer les données fournies par des services à travers des mashups. / We propose an approach for adding non-functional properties (exception handling, atomicity, security, persistence) to services' coordinations. The approach is based on an Active Policy Model (AP Model) for representing services' coordinations with non-functional properties as a collection of types. In our model, a services' coordination is represented as a workflow composed of an ordered set of activities, each activity in charge of implementing a call to a service' operation. We use the type Activity for representing a workflow and its components (i.e., the workflow' activities and the order among them). A non-functional property is represented as one or several Active Policy types, each policy composed of a set of event-condition-action rules in charge of implementing an aspect of the property. Instances of active policy and activity types are considered in the model as entities that can be executed. We use the Execution Unit type for representing them as entities that go through a series of states at runtime. When an active policy is associated to one or several execution units, its rules verify whether each unit respects the implemented non-functional property by evaluating their conditions over their execution unit state, and when the property is not verified, the rules execute their actions for enforcing the property at runtime. We also proposed a proof of concept Active Policy Execution Engine for executing an active policy oriented workflow modelled using our AP Model. The engine implements an execution model that determines how AP, Rule and Activity instances interact among each other for adding non-functional properties (NFPs) to a workflow at execution time. We validated the AP Model and the Active Policy Execution Engine by defining active policy types for addressing exception handling, atomicity, state management, persistency and authentication properties. These active policy types were used for implementing reliable service oriented applications, and mashups for integrating data from services.

The increasing size of high performance computing systems and the associated increase in the volume of generated data, has resulted in an I/O bottleneck for these applications. This bottleneck is further exacerbated by the imbalance in the growth of processing capability compared to storage capability, due mainly to the power and cost requirements of scaling the storage. This thesis introduces data services, a new abstraction which provides significant benefits for data intensive applications. Data services combine low overhead data movement with flexible placement of data manipulation operations, to address the I/O challenges of leadership class scientific applications. The impact of asynchronous data movement on application runtime is minimized by utilizing novel server side data movement schedulers to avoid contention related jitter in application communication. Additionally, the JITStager component is presented. Utilizing dynamic code generation and flexible code placement, the JITStager allows data services to be executed as a pipeline extending from the application to storage. It is shown in this thesis that data services can add new functionality to the application without having an significant negative impact on performance.

Data science employment opportunities of varied complexity and environment are in growing demand across the globe. Data science as a discipline potentially offers a wealth of jobs to prospective employees, while traditional information science-based roles continue to decrease as budgets get cut across the U.S. Since data is related closely to information historically, this research will explore the education of U.S. iSchool professionals and compare it to traditional data science roles being advertised within the job market. Through a combination of latent semantic analysis of over 1600 job postings and iSchool course documentation, it is our aim to explore the intersection of library and information science and data science. Hopefully these research findings will guide future directions for library and information science professionals into data science driven roles, while also examining and highlighting the data science techniques currently driven by the education of iSchool professionals. In addition, it is our aim to understand how data science could benefit from a mutually symbiotic relationship with the field of information science as statistically data scientists spend far too much time working on data preparation and not nearly enough time conducting scientific inquiry. The results of this examination will potentially guide future directions of iSchool students and professionals towards more cooperative data science roles and guide future research into the intersection between iSchools and data science and possibilities for partnership.

Spaceborne Remote Sensing, the observation of the features on ground from orbiting satellites is one of the expanding business areas of international space business, attracting the analysis of the markets and the growth patterns in the international arena. Whatever be the product/service, it is important for the market players to know the market characteristics to draw up marketing strategy and defining marketing mix, which becomes important for Remote Sensing services as well without an exception. The market for Remote Sensing data and services is influenced by a multitude of factors, which may seem unrelated at the outset. These may be Socio-Economic and Natural Resources related variables in addition to the expected product/market related variables. Hence the research work to model the market dynamics is taken up with the objectives of: 1)To identify factors influencing the market potential for Remote Sensing Data services 2)To model the relationship between these factors and the market potential 3)To validate and verify the model developed The analysis was carried out with 82 valid responses from respondents spread across 17 countries where Remote Sensing market is prevailing. A broad spectrum of economic, political, social, natural, and geographic variables believed to be influencing the Remote Sensing requirements were enumerated and validated through pilot study. Raw values for the variables were obtained from secondary sources and relative importance of these variables was determined through primary survey. The variables were synthesised into factors and captured into a regression model to arrive at the market potential index. This is statistically validated and also verified with market potential of a known country market, India. The analysis and results revealed that, market dynamics influencing market potential for Spaceborne Remote Sensing Data can be synthesised as the following three factors, explaining 80% of the variance in the market potential. (i)Politico Developmental factor-variables border related (Number of border countries, Border length, level of Border Disputes, Energy Production & Consumption, Economy and Information & Communication Infrastructure) (ii)Geo-Natural factor: represented by extent of Natural Resources, Energy Reserves etc. (iii)Market/Product related factor variables like product features, level of competition, availability of substitutes, current market position, and growth rate etc. The emergence of variables representing the developmental aspects on one side and political aspects on the other, as important factor is the outcome formally brought to recognition by the study. This is manifesting the growing importance of developmental activities and theatres of war as important additional market drivers in high and very-high resolution data services across the globe, in addition to the traditional applications like natural resources monitoring and management.

This thesis is concerned with analysis of the mobile operators in the Czech Republic. Introduction part of thesis is devoted to the explanation of the key indicators which are used by mobile operators to analyze the rate of customer churn, or average revenue per user, or the customer value. Very importat is also a comparison of price consumption baskets in the Czech Republic with the other countries. The situation has changed dramatically offering unlimited tariffs in the already fully saturated market. In the practical part of this thesis I have focused on a detailed analysis of customers data in Vodafone company a I have reviewed the current offer of affordable tariffs and data services. I tried to implement new tariffs based on a real information of all customers - individuals over a period of 3 months. The aim of the thesis is to propose tariffs and services for customers, which would maintain the current customer base or which would increase the base slightly.

Cette thèse s’intéresse aux données de la recherche, dans un contexte d’incitation croissante à leur ouverture. Les données de la recherche sont des informations collectées par les scientifiques dans la perspective d’être utilisées comme preuves d’une théorie scientifique. Il s’agit d’une notion complexe à définir, car contextuelle. Depuis les années 2000, le libre accès aux données occupe une place de plus en plus stratégique dans les politiques de recherche. Ces enjeux ont été relayés par des professions intermédiaires, qui ont développé des services dédiés, destinés à accompagner les chercheurs dans l’application des recommandations de gestion et d’ouverture. La thèse interroge le lien entre idéologie de l’ouverture et pratiques de recherche. Quelles formes de gestion et de partage des données existent dans les communautés de recherche et par quoi sont-elles motivées ? Quelle place les chercheurs accordent-ils à l’offre de services issue des politiques de gestion et d’ouverture des données ? Pour tenter d’y répondre, 57 entretiens ont été réalisés avec des chercheurs de l’Université de Strasbourg dans différentes disciplines. L’enquête révèle une très grande variété de pratiques de gestion et de partage de données. Un des points mis en évidence est que, dans la logique scientifique, le partage des données répond un besoin. Il fait partie intégrante de la stratégie du chercheur, dont l’objectif est avant tout de préserver ses intérêts professionnels. Les données s’inscrivent donc dans un cycle de crédibilité, qui leur confère à la fois une valeur d’usage (pour la production de nouvelles publications) et une valeur d’échange (en tant que monnaie d’échange dans le cadre de collaborations avec des partenaires). L’enquête montre également que les services développés dans un contexte d’ouverture des données correspondent pour une faible partie à ceux qu’utilisent les chercheurs. L’une des hypothèses émises est que l’offre de services arrive trop tôt pour rencontrer les besoins des chercheurs. L’évaluation et la reconnaissance des activités scientifiques étant principalement fondées sur la publication d’articles et d’ouvrages, la gestion et l’ouverture des données ne sont pas considérées comme prioritaires par les chercheurs. La seconde hypothèse avancée est que les services d’ouverture des données sont proposés par des acteurs relativement éloignés des communautés de recherche. Les chercheurs sont davantage influencés par des réseaux spécifiques à leurs champs de recherche (revues, infrastructures…). Ces résultats invitent finalement à reconsidérer la question de la médiation dans l’ouverture des données scientifiques. / The thesis investigates research data, as there is a growing demand for opening them. Research data are information that is collected by scientists in order to be used as evidence for theories. It is a complex, contextual notion. Since the 2000s, open access to scientific data has become a strategic axis of research policies. These policies has been relayed by third actors, who developed services dedicated to support researchers with data management and sharing.The thesis questions the relationship between the ideology of openness and the research practices. Which kinds of data management and sharing practices already exist in research communities? What drives them? Do scientists rely on research data services? Fifty-seven interviews were conducted with researchers from the University of Strasbourg in many disciplines. The survey identifies a myriad of different data management and sharing practices. It appears that data sharing is embedded in the researcher’s strategy: his main goal is to protect his professional interests. Thus, research data are part of a credibility cycle, in which they get both use value (for new publications) and exchange value (as they are traded for other valuable resources). The survey also shows that researchers rarely use the services developed in a context of openness. Two explanations can be put forward. (1) The service offer comes too early to reach researchers’ needs. Currently, data management and sharing are not within researchers’ priorities. The priority is publishing, which is defined as source of reward and recognition of the scientific activities. (2) Data management services are offered by actors outside the research communities. But scientists seem to be more influenced by internal networks, close to their research topics (like journals, infrastructures…). These results prompt us to reconsider the mediation between scientific communities and open research data policies.

This study applies the knowledge, skills, and abilities (KSA) framework for eScience professionals to data service positions in academic libraries. Understanding the KSAs needed to provide data services is of crucial concern. The current study looks at KSAs of data professionals working in the United States academic libraries. An exploratory sequential mixed method design was adopted to discover the KSAs. The study was divided into two phases, a qualitative content analysis of 260 job advertisements for data professionals for Phase 1, and distribution of a self-administered online survey to data professionals working in academic libraries research data services (RDS) for Phase 2. The discovery of the KSAs from the content analysis of 260 job ads and the survey results from 167 data professionals were analyzed separately, and then Spearman rank order correlation was conducted in order to triangulate the data and compare results. The results from the study provide evidence on what hiring managers seek through job advertisements in terms of KSAs and which KSAs data professionals find to be important for working in RDS. The Spearman rank order correlation found strong agreement between job advertisement KSAs and data professionals perceptions of the KSAs.

Abstract in English, Afrikaans and Zulu / The aim of this study was to determine the key factors which explain the adoption of mobile data services in South Africa and incorporate them into an explanatory framework that can be used to indicate whether a new mobile data service that is introduced into the South African Mobile Market will be successful. Additionally, it shows that the current usage of mobile data services in South Africa supports the framework. The research used a design based on a mixed methods methodology. The research was conducted using a literature survey and two research phases. The literature survey was an extensive review of communication technology adoption frameworks and all the possible factors of adoption. This literature survey was also used to develop the discussion document which was used as the guidelines for the Phase 1 interviews. The interviews were with experts in the telecommunications market in South Africa and were used to determine the key drivers of adoption and possible moderating factors of mobile data services. From those interviews a preliminary model was proposed. Phase 2 was a market survey which tested the framework and moderating factors in different metropolitan areas as well as examined the current usage of mobile data services. The statistical analysis used in determining the framework incorporated Descriptive Statistics, Factor Analysis, using Principal Axis Factoring, and Structural Equation Modelling. The framework, which was based on the Technology Acceptance Model, introduced two new constructs. The first new construct was the Mobile Service Providers Marketing Tactics and the second was Social Pressure and Aspirational Value. The two constructs combined can be used to explain why mobile service providers with low cost strategies are not successful. Demographic factors such as Age were found to influence the adoption as well as Socio-economic factors such as Income. Personal factors such as Technical Knowledge, Ability and Skills and Attitude towards Technology were also found to influence adoption. Geographical location was found to be the most important moderating factor. The current mobile data usage supported the framework. The results of data usage highlighted the fact that nearly 50% of the study’s participants spent over 4 hours per day interacting with their mobile devices and that the participants from Gauteng spent more than 2½ times on mobile services than those in the other metropolitan municipalities of Cape Town and eThekwini / Die doel van hierdie studie was om die sleutelfaktore te bepaal wat die aanneming van selfoondatadienste in Suid-Afrika verklaar, en hulle te verenig in ʼn raamwerk waarmee aangetoon kan word of ʼn nuwe selfoondatadiens die Suid-Afrikaanse selfoondatamark suksesvol sal betree. Afgesien hiervan dui hierdie studie aan dat die huidige gebruik van selfoondatadienste in Suid-Afrika die geldigheid van die voorgestelde raamwerk bevestig. ʼn Mengsel van metodes is in die navorsing toegepas. Die navorsing het uit ʼn literatuuroorsig en twee navorsingsfases bestaan. Die literatuuroorsig het ʼn deurtastende oorsig van die raamwerke vir die aanneming van kommunikasietegnologie en alle moontlike faktore daarvoor behels. Uit die literatuuroorsig is die besprekingsdokument saamgestel wat as riglyn gedien het vir die onderhoude in fase 1 van die navorsing. Onderhoude is met kundiges uit die Suid-Afrikaanse telekommunikasiebedryf gevoer om die belangrikste dryfvere vir aanneming en moontlike remfaktore ten opsigte van selfoondatadienste te bepaal. ʼn Voorlopige model is op grond van die onderhoude aangebied. Fase 2 het ʼn markopname behels waarin die raamwerk en remfaktore in verskeie metropolitaanse gebiede getoets en die huidige gebruik van selfoondatadienste ondersoek is. Die statistiese analise waarop die raamwerk berus, het bestaan uit beskrywende statistiek, faktorontleding met behulp van hoofasfaktorering, en struktuurvergelykingsmodellering. Twee nuwe konstrukte het uit die raamwerk wat op die tegnologieaanvaardingsmodel gegrond was, gespruit. Die eerste was selfoondiensverskaffers se bemarkingstrategieë en die tweede sosiale druk en aspirerende waarde. Saam verklaar die twee konstrukte waarom selfoondiensverskaffers wat laekostestrategieë volg, misluk. Daar is bevind dat demografiese faktore soos ouderdom, en sosiaal-ekonomiese faktore soos inkomste, aanneming beïnvloed. Volgens die bevindings word aanneming eweneens bepaal deur persoonlike faktore soos tegniese kennis, vermoë en vaardighede asook houding jeens tegnologie. Daar is bevind dat geografiese ligging die belangrikste remfaktor is. Die huidige selfoondatagebruik bevestig die geldigheid van die raamwerk. Die uitslag van datagebruik benadruk die feit dat byna 50% van die deelnemers aan die studie langer as vier uur per dag met hulle selfone doenig is, en dat deelnemers in Gauteng twee en half keer meer as deelnemers in die metropolitaanse munisipaliteite van Kaapstad en eThekwini aan selfoondienste bestee. / Inhloso yalolu cwaningo bekuwukuthola izinto ezibalulekile ezichaza ukwamukelwa kwezinsiza zamadatha amaselula eNingizimu Afrika nokuzifaka ohlakeni oluchazayo olungasetshenziswa ukukhombisa ukuthi ngabe insiza entsha yedatha ethulwe emakethe yamaselula eNingizimu Afrika izophumelela. Ngaphezu kwalokho, kukhombisa ukuthi ukusetshenziswa kwamanje kwezinsiza zedatha yamaselula eNingizimu Afrika kuyalusekela uhlaka. Ucwaningo lusebenzise umklamu osuselwe ezindleleni zokwenza ezixubile. Ucwaningo lwenziwe kusetshenziswa ucwaningo lwezincwadi kanye nezigaba ezimbili zocwaningo. Ucwaningo lwezincwadi belungukubuyekezwa okubanzi kokwamukela kwenzinhlaka zobuchwepheshe kwezokuxhumana kanye nazo zonke izici ezingenzeka ukuthi zamukelwe. Lolu cwaningo lwezincwadi lusetshenzisiwe futhi ukuthuthukisa umqulu wezingxoxo osetshenziswe njengezinkombandlela zokuxoxisana nabantu besigaba 1. Izingxoxo ebezinazo bezingochwepheshe emakethe yezokuxhumana eNingizimu Afrika futhi zasetshenziselwa ukuthola izidingongqangi zokwamukelwa kanye nezici zokulingisanisa kwezinsiza zamadatha amaselula. Kulezo zingxoxo, kuphakanyiswe isifaniselo sokuqala. Isigaba 2 bekungukucwaningo lwezimakethe oluhlola uhlaka kanye nezici zokulinganisa ezindaweni ezahlukahlukene zedolobha kanye nokuhlola ukusetshenziswa kwamanje kwezinsiza zedatha yamaselula. Ukuhlaziywa kwezibalo okusetshenzisiwe ekunqumeni uhlaka lufake izibalo ezichazayo, ukuhlaziya izakhi, kusetshenziswa izisekelo eziyinhloko zokuphamba, kanye nesifaniselo sokulinganisa esihlelekile. Uhlaka, olwalususelwa kusifaniselo sokwamukela ubuchwepheshe, lwethule izakhiwo ezimbili ezintsha. Isakhiwo esisha sokuqala kwakungamacebo wokukhangisa wabahlinzeki bezinsiza zamaselula futhi owesibili kwakuyingcindezi yomphakathi nenani lesifiso. Lezi zakhiwo zombili ezihlanganisiwe zingasetshenziswa ukuchaza ukuthi kungani abahlinzeki bezinsizakalo zamaselula abanamasu ezindleko eziphansi bengaphumeleli. Izici ezithinta abantu ezinjengobudala zitholakele ukuthonya ukwamukelwa kanye nezici zezomnotho nezenhlalo ezifana nomholo. Izici ezithinta umuntu ngqo ezifana nolwazi lwezobuchwepheshe, ukwazi ukwenza kanye namakhono, nesimo sengqondo maqondana nezobuchwepheshe zatholakala zithonya ukwamukelwa. Indawo yendawo yatholakala njengesici sokulinganisa esibaluleke kakhulu. Ukusetshenziswa kwedatha yeselula yamanje kusekele uhlaka. Imiphumela yokusetshenziswa kwedatha igcizelele iqiniso lokuthi bacishe babe ngama-50% ababambe iqhaza kulolu cwaningo bachithe isikhathi esingaphezu kwamahora amane ngosuku bexhumana nemishini yabo yamaselula nokuthi ababambiqhaza abavela eGauteng bachithe izikhathi ezingaphezu kwezingu-2½ kuzinsiza zamaselula kunalabo bakwamanye amadolobha amakhulu aseKapa naseThekwini. / Business Management / D.B.L.