The purpose of this study is to examine the effect of experience staging techniques (personalization through co-creation and multisensory stimuli) on engagement level. This study also explores the possible contribution of experience staging techniques as practical tools that recreation professionals can use to better engage participants in recreation activities and events. A 2-way univariate ANOVA revealed no significant relationship between the use of co-creative and multisensory stimulating techniques and engagement levels in participants (F (3,200) = .263, p = .826, partial η2 = .004). Practical applications for recreation professionals and further research opportunities are discussed.

<div>Federated learning (FL) is a decentralized privacy-preserving learning technique in which clients learn a joint collaborative model through a central aggregator without sharing their data. In this setting, all clients learn a single common predictor (FedAvg), which does not generalize well on each client's local data due to the statistical data heterogeneity among clients. In this paper, we address this problem with PersFL, a discrete two-stage personalized learning algorithm. In the first stage, PersFL finds the optimal teacher model of each client during the FL training phase. In the second stage, PersFL distills the useful knowledge from optimal teachers into each user's local model. The teacher model provides each client with some rich, high-level representation that a client can easily adapt to its local model, which overcomes the statistical heterogeneity present at different clients. We evaluate PersFL on CIFAR-10 and MNIST datasets using three data-splitting strategies to control the diversity between clients' data distributions.</div><div><br></div><div>We empirically show that PersFL outperforms FedAvg and three state-of-the-art personalization methods, pFedMe, Per-FedAvg and FedPer on majority data-splits with minimal communication cost. Further, we study the performance of PersFL on different distillation objectives, how this performance is affected by the equitable notion of fairness among clients, and the number of required communication rounds. We also build an evaluation framework with the following modules: Data Generator, Federated Model Generation, and Evaluation Metrics. We introduce new metrics for the domain of personalized FL, and split these metrics into two perspectives: Performance, and Fairness. We analyze the performance of all the personalized algorithms by applying these metrics to answer the following questions: Which personalization algorithm performs the best in terms of accuracy across all the users?, and Which personalization algorithm is the fairest amongst all of them? Finally, we make the code for this work available at https://tinyurl.com/1hp9ywfa for public use and validation.</div>

There is a great need for digital libraries that are flexible, scalable, and reliable. Few such systems exist. Little is known about how to build them. This thesis addresses these problems by enhancing a prototype digital library system with the aim of making it more flexible, scalable, and reliable. We hypothesize that: 1) adding a new (user information) layer and maintaining weak coupling in the design of a digital library system can help achieve system flexibility; 2) optimizing network connection usage and facilitating distribution of computation and disk operations in system design can help achieve system scalability; and 3) applying good software processes can help university students produce a very reliable system. Approaches based on the above hypothesis were used in the project of Reengineering the MARIAN System in Java. The results of the project and experiments verified the correctness of the hypothesis. The results of this thesis may help inform future digital library design and implementation projects to produce flexible, scalable, and reliable systems. / Master of Science

Depuis le « Plan Informatique Pour Tous » de 1985, les technologies numériques ne cessent d'occuper une place grandissante dans l'enseignement : manuels numériques, logiciels de géométrie dynamique, learning games, e-learning, blended learning, MOOC, classes inversées, robotique éducative, etc.L'ambition de nos travaux est de montrer que certaines de ces technologies peuvent contribuer à améliorer les apprentissages, en dynamisant les contenus, en accentuant la motivation des étudiants, en proposant des dispositifs adaptés à la formation à distance, en personnalisant les parcours pédagogiques.Les enjeux autour de ces questions sont importants. La nécessité de motiver les étudiants et de personnaliser les apprentissages apparaît de plus en plus clairement. Ce sont des atouts majeurs pour lutter contre le décrochage scolaire et pour l'égalité des chances.Objectifs de nos travaux antérieurs à 2011 :— Ludifier et animer des contenus afin de les rendre plus motivants et plus explicites.— Visualiser des concepts en manipulant des objets numériques.— Virtualiser des objets d'apprentissage pour s'affranchir de contraintes matérielles afin de faire travailler des méthodes, de dépasser des difficultés de manipulation et des situations de handicap.— Fournir des outils d'interactivité, de visualisation, de calcul formel et de géométrie pour des environnements informatiques d'apprentissage (plateformes d'enseignement à distance, logiciels).— Fournir des outils de monitoring des activités des utilisateurs afin de suivre au mieux leur progression, afin de pouvoir les suivre au plus près dans leurs cheminements, de leur fournir des retours adaptés et des parcours personnalisés, de les rendre plus autonomes.— Expérimenter des objets à la fois numériques et tangibles tels que les robots pour évaluer leur impact dans les apprentissages.— Repenser les manuels scolaires en les accompagnant de dispositifs numériques.Ces travaux ont trouvé un prolongement ciblé, fortement ancré recherche, dans des travaux plus récents.Objectifs de nos travaux postérieurs à 2011 :— Optimiser et personnaliser en profondeur les apprentissages en faisant appel à l'intelligence artificielle et à des algorithmes de machine learning.— Introduire des objets tangibles, tels que les robots, que les élèves peuvent manipuler, voire programmer, pour éclairer différemment les apprentissages et proposer une approche concrète pour construire de nouveaux concepts. / Since the "Plan Informatique Pour Tous" in 1985, digital technologies occupy an increasingly importance in education: digital textbooks, dynamic geometry software, learning games, e-learning, blended learning, MOOC, flipped classrooms, educational robotics, etc.The aim of our work is to show that some of these technologies can contribute to improve learning, boosting learning contents, emphasizing student motivation by proposing devices suitable for distance learning and personalizing learning paths.The stakes of these issues are important. The need to motivate students and personalize learning is more and more crucial. These are major assets to reduce dropout and promote equal opportunities.Objectives of our work before 2011:- Gamify contents to make them more motivating.- Visualize concepts by using digital objects.- Virtualize learning objects in order to reduce physical constraints to work methods, to overcome handling difficulties and disability situations.- Provide tools for interactivity, visualization, computer algebra and geometry for computer environments learning (distance learning platforms, software).- Provide tools for monitoring user activity in order to better track their progress, to follow them with precision, to making them more autonomous.- Experiment with objects both digital and tangible such as robots, to assess their impact in learning.- Build new textbooks by accompanying them with digital devices.This work was continued in recent and more research-driven work.Objectives of our work from 2011:- Optimize and personalize learning by using artificial intelligence and machine learning algorithms.- Use tangible objects such as robots, that students can manipulate and program, to approach learning differently to provide concrete environment to build new concepts.

La personnalisation de l’apprentissage dans les environnements informatiques pour l’apprentissage humain (EIAH) est un sujet de recherche qui est traité depuis de nombreuses années. Avec l’arrivée des cours en ligne ouverts et massifs (MOOC), la question de la personnalisation se pose de façon encore plus cruciale et de nouveaux défis se présentent aux chercheurs. En effet, le même MOOC peut être suivi par des milliers d’apprenants ayant des profils hétérogènes (connaissances, niveaux éducatif, objectifs, etc). Il devient donc nécessaire de tenir compte de cette hétérogénéité en présentant aux apprenants des contenus éducatifs adaptés à leurs profils afin qu’ils tirent parti au mieux du MOOC.D’un autre côté, de plus en plus de ressources éducatives libres (REL) sont partagées sur le web. Il est important de pouvoir réutiliser ces REL dans un contexte différent de celui pour lequel elles ont été créées. En effet, produire des REL de qualité est une activité coûteuse en temps et la rentabilisation des REL passe par leur réutilisation.Pour faciliter la découverte des REL, des schémas de métadonnées sont utilisés pour décrire les REL.Cependant, l’utilisation de ces schémas a amené à des entrepôts isolés de descriptions hétérogènes et qui ne sont pas interopérables. Afin de régler ce problème, une solution adoptée dans la littérature consiste à appliquer les principes des données ouvertes et liées (LOD) aux descriptions des REL.Dans le cadre de cette thèse, nous nous intéressons à la personnalisation des MOOC et à la réutilisation des REL.Nous proposons un système de recommandation qui fournit à un apprenant en train de suivre un MOOC des ressources externes qui sont des REL adaptées à son profil, tout en respectant les spécificités du MOOC suivi.Pour sélectionner les REL, nous nous intéressons à celles qui possèdent des descriptions insérées dans les LOD, stockées dans des entrepôts accessibles sur le web et offrant des moyens d’accès standardisés. Notre système de recommandation est implémenté dans une plateforme de MOOC, Open edX et il est évalué en utilisant une plateforme de micro-tâches. / For many years now, personalization in TEL is a major subject of intensive research. With the spreading of Massive Open Online Courses (MOOC), the personalization issue becomes more acute. Actually, any MOOC can be followed by thousands of learners with different educational levels, learning styles, preferences, etc. So, it is necessary to present pedagogical contents taking into account their heterogeneous profiles so that they can maximize their benefit from following the MOOC.At the same time, the amount of Open Educational Resources (OER) available on the web is permanently growing. These OERs have to be reused in contexts different from the initial ones for which they were created.Indeed, producing quality OER is costly and requires a lot of time. Then, different metadata schemas are used to describe OER. However, the use of these schemas has led to isolated repositories of heterogeneous descriptions which are not interoperable. In order to address this problem, a solution adopted in the literature is to apply Linked Open Principles (LOD) to OER descriptions.In this thesis, we are interested in MOOC personalization and OER reuse. We design a recommendation technique which computes a set of OERs adapted to the profile of a learner attending some MOOC. The recommended OER are also adapted to the MOOC specificities. In order to find OER, we are interested in those who have metadata respecting LOD principles and stored in repositories available on the web and offering standardized means of access. Our recommender system is implemented in the MOOC platform Open edX and assessed using a micro jobs platform.

Syfte: Studien ämnar att skapa en djupare förståelse kring hur personalization-privacy paradoxen behandlas i hotellbranschen utifrån en individanpassad marknadsföring. Metod: Studiens metod innefattar en kvalitativ undersökningsmetod där semistrukturerade intervjuer genomförts med åtta respondenter från hotellbranschen som driver marknadsföringen. De semistrukturerade intervjuerna bygger på att studera marknadsförares inställning till personalization-privacy paradoxen utifrån en individanpassad marknadsföring. Resultat och slutsats: Vår tolkning av intervjusvaren indikerar att den individanpassade marknadsföringen starkt uppskattas och efterfrågas av marknadsförare på hotellbranschen. Marknadsföringsmetoden ses vara mer effektiv än allmän marknadsföring då det är enklare att hitta rätt målgrupp samt att det är en större möjlighet att kunden accepterar ett erbjudande. Integritetsparadoxen ses som ett stort dilemma där marknadsförare ska följa regelverket men även sin moraliska kompass. Syftet ska vara att skapa en god personlig service till kunderna och inte nyttja informationen till ens egna vinstintressen. En gräns kan dras där hur långt marknadsförare kan motivera varför kundens information samlas in och hur den kommer att användas. Uppsatsens bidrag: Detta arbete bidrar till att skapa förståelse kring marknadsförares syn på Personalization-privacy paradox i hotellbranschen utifrån en individanpassad marknadsföring. Förslag till vidare forskning: Till vidare forskning hade det varit intressant med studiens resultat göra en ytterligare forskning som jämför konsumenternas syn på personalization-privacy paradoxen i hotellbranschen utifrån en individanpassad marknadsföring. / Purpose: The study aims to create a deeper understanding of how the personalization-privacy paradox is treated in the hotel industry based on individualized marketing. Method: The method includes a qualitative survey method where semi-structured interviews were conducted with eight respondents from the hotel industry who drive the marketing. The semi-structured interviews are based on studying marketers' attitudes towards personalization privacy based on individualized marketing. Results and conclusion: Our interpretation of the interview results indicates that individualized marketing is highly appreciated and in demand by marketers in the hotel industry. The marketing method is seen to be more effective than general marketing as it is easier to find the right target group and there is a greater possibility that the customer accepts an offer. The integrity paradox is seen as a major dilemma where marketers must follow the regulations but also their moral compass. The purpose should be to create a good personal service to customers and not use the information for their own profit interests. A limit is drawn where how far can motivate why the customer is collected and how it will be used. Contribution: This study helps to create a deeper understanding of the marketers' view of the Personalization-privacy paradox in the hotel industry based on individualized marketing. Proposal for further research: For further research, it would have been interesting with the result of the study to do further research that compares the consumers view on the personalization-privacy paradox in the hotel industry based on an individualized marketing.

No description available.

In this thesis, the process of developing an application for segmenting customers with the use of machine learning is described. The project was carried out at a company which provides a booking platform for beauty and health services. Data about customers were analyzed and processed in order to train two classification models able to segment customers into three different customer groups. The performance of the two models, a Logistic Regression model and a Support Vector Classifier, were evaluated with different numbers of features and compared to classifications made by human experts working at the company. The results shows that the logistic regression model achieved an accuracy of 71% when classifying users into the three groups, which was more accurate than the experts manual classification. A web API where the model is provided has been developed and presented to the company. The results of the study showed that machine learning is a useful technique for performing customer segmentation based on behavioral data. Even in the case where the classes are not naturally divisible, the application provides valuable insights on user behaviour that can help the company become more data-driven.

Background: As personalization has become a common CRM strategy for companies to create valuable relationships with customers, users are receiving an increased amount of personalized communication, further research is needed on the influence of content personalization in specific channels, to improve customer engagement.  Purpose: This paper seeks to analyse the influence of eCRM content personalization strategies on the consumer engagement performance of email marketing campaigns building upon existing knowledge about the benefits and opportunities of personalization strategies.  Method: The selected method is quantitative, using A/B testing, based on the comparison of different variations on two identical segments where the only differentiated element is the one being evaluated. Two controlled experiments evaluating subject line and image personalization are performed evaluating four metric of email marketing performance: open rate, click-through rate, conversion rate and unsubscribe rate.  Conclusion: The results of the two controlled experiments performed for this research, subject line personalization and image personalization, complement past literature on content personalization strategies (Bertrand et al., 2010; Carlota Rocha et al., 2023; Munz et al., 2020; Sahni et al., 2018) by demonstrating the statistically significant positive influence of content personalization on email marketing performance and reinforcing the importance of familiarity with the brand or product on reducing the probability of bad outcomes.

Elders usually experience high levels of loneliness in the society as they have retired from work and less of a center within the family. There are many interaction design work that target elder’s loneliness and aim to solve the problem by connecting elders with other people through distance. Digitalization’s connectedness is a helpful factor that can combat elder’s communication issues. At the same time, there are many already existing digital services that embed communication functions such as email, texting, and video call. Still, elders find themselves have difficulties using existing technology. The thesis focuses on why current digital technologies cannot help elders connect and communicate efficiently and how to re-design to help elders feel their willingness to communicate by current technologies. Empirical research in the thesis project has shown that elders lack trust towards digital systems due to lack of transparency and findings of specialized contexts. The project is in two parts: qualitative research and prototyping. Through qualitative research, the thesis shows that there is a lack of transparency and control from digital applications that elders experience. Prototyping outcomes suggests enabling personalization within digital services for elders by implementing slow user interactions and simple graphic elements.