This study investigates the growing integration of Artificial Intelligence (AI) in educational settings, specifically focusing on detecting AI-generated content in students’ English essays. As AI technologies like ChatGPT and Gemini become more prevalent, understanding their impact on education is crucial. This research aims to identify the linguistic features that lead English as a Foreign Language (EFL) teachers to suspect AI involvement in student work. By conducting semi-structured interviews with eight EFL teachers from lower upper secondary and high schools, the study examines their experiences and perspectives. Using the Technological Pedagogical Content Knowledge (TPCK) framework, the study analyzes the crossing of technology, pedagogy, and content knowledge, highlighting the opportunities and challenges AI presents in contemporary education. The findings show the dual role of AI as both a beneficial tool for improving learning and a challenge to maintaining academic integrity. Despite the limitations, such as the evolving nature of AI, the research highlights the need for teachers to balance the benefits of AI with preserving authentic student work. Future research directions include exploring more effective AI detection methods and understanding the long-term impact of AI on students’ critical thinking skills.

Objective The goal of this retrospective study is the evaluation of risk factors for postoperative neurological deficits after petroclival meningioma (PCM) surgery with special focus on standard craniotomies. Materials and methods One-hundred-fifty-eight patients were included in the study, of which 133 patients suffered from primary and 25 from recurrent PCM. All patients were operated on and evaluated concerning age, tumor size, histology, pre- and postoperative cranial nerve (CN) deficits, morbidity, mortality, and surgical complications. Tumor-specific features—e.g., consistency, surface, arachnoid cleavage, and location—were set in a four-grade classification system that was used to evaluate the risk of CN deficits and tumor resectability. Results After primary tumor resection, new CN deficits occurred in 27.3% of patients. Preoperative ataxia improved in 25%, whereas 10% developed new ataxia. Gross total resection (GTR) was achieved in 59.4%. The morbidity rate, including hemiparesis, shunt-dependence, postop-hemorrhage, and tracheostomy was 22.6% and the mortality rate was 2.3%. In recurrent PCM surgery, CN deficits occurred in 16%. GTR could be achieved in three cases. Minor complications occurred in 20%. By applying the proposed new classification system to patients operated via standard craniotomies, the best outcome was observed in type I tumor patients (soft tumor consistency, smooth surface, plane arachnoid cleavage, and unilateral localization) with GTR in 78.7% (p < 0.001) and 11.9% new CN deficits (p = 0.006). Conclusion Standard craniotomies as the retrosigmoid or subtemporal/pterional approaches are often used for the resection of PCMs. Whether these approaches are sufficient for GTR—and avoidance of new neurological deficits—depends mainly on the localization and intrinsic tumor-specific features.

Esta tesis se centra en la construcción y usos de los modelos metabólicos a escala genómica para obtener biocombustibles de manera eficiente, como etanol e hidrógeno. Como organismo objetivo, se ha elegido a la cianobacteria Synechocystis sp. PCC6803. Este organismo ha sido estudiado como una potencial plataforma de producción alimentada por fotones, dada su capacidad de crecer solamente a partir de dióxido de carbono y fotones. Esta tesis versa acerca de los métodos para modelar, analizar, estimar y predecir el comportamiento del metabolismo de las células. La principal meta es extraer conocimiento de los diferentes aspectos biológicos de un organismo con el fin de utilizarlo para un objetivo industrial pertinente. Esta tesis ha sido estructurada en capítulos organizados de acuerdo con las sucesivas tareas que terminan con la construcción de una célula in silico que se comporta, idealmente, como la que está basada en el carbono. Este proceso suele comenzar con los archivos de anotación del genoma y termina con un modelo metabólico a escala genómica capaz de integrar datos -ómicos. El primer objetivo de la presente tesis es la reconstrucción de un modelo del metabolismo de esta cianobacteria que tenga en cuenta todas las reacciones presentes en la misma. Esta reconstrucción tenía que ser lo suficientemente flexible como para permitir el crecimiento en las distintas condiciones ambientales bajo las cuales este organismo crece en la naturaleza, así como permitir la integración de diferentes niveles de información biológica. Una vez que se cumplió este requisito, se pudieron simular variaciones ambientales y estudiar sus efectos desde una perspectiva de sistema. Se han estudiado hasta cinco diferentes condiciones de crecimiento en este modelo metabólico y sus diferencias han sido evaluadas. La siguiente tarea fue definir estrategias de producción para sopesar la viabilidad de este organismo como una plataforma de producción. Se simularon perturbaciones genéticas para e / This thesis is focused on the construction and uses of genome-scale metabolic models to efficiently obtain biofuels, such as ethanol and hydrogen. As a target organism, cyanobacterium Synechocystis sp. PCC6803 was chosen. This organism has been studied as a potential photon-fuelled production platform, for its ability to grow only from carbon dioxide, water and photons. This dissertation verses about methods to model, analyse, estimate and predict the metabolic behaviour of cells. Principal goal is to extract knowledge from the different biological aspects of an organism in order to use it for an industrial relevant objective. This dissertation has been structured in chapters accordingly organized as the successive tasks that end up building an in silico cell that behaves as the carbon-based one. This process usually starts with the genome annotation files and ends up with a genome-scale metabolic model able to integrate ¿omics data. First objective of present thesis is to reconstruct a model of this cyanobacteria¿s metabolism that accounts for all the reactions present in it. This reconstruction had to be flexible enough as to allow growth under the different environmental conditions under which this organism grows in nature as well as to allow the integration of different levels of biological information. Once this requisite was met, environmental variations could be simulated and their effect studied under a system-wide perspective. Up to five different growth conditions were simulated on this metabolic model and differences were evaluated. Following assignment was to define production strategies to weigh this organism¿s viability as a production platform. Genetic perturbations were simulated to design strains with an enhanced production of three industrially-relevant metabolites: succinate, ethanol and hydrogen. Resulting sets of genetic modifications for the overproduction of those metabolites are, thus, proposed. Moreover, functional reactions couplings were studied and weighted to their metabolite production importance. Finally, genome-scale metabolic models allow establishing integrative approaches to include different types of data that help to find regulatory hotspots that can be targets of genetic modification. Such regulatory hubs were identified upon light/dark shifts and general metabolism operational principles inferred. All along this process, blind spots in Synechocystis sp. PCC6803 metabolism, and more importantly, blind spots in our understanding of it, are revealed. Overall, the work presented in this thesis unveils the industrial capabilities of cyanobacterium Synechocystis sp. PCC6803 to evolve interesting metabolites as a clean production platform. / Esta tesis es centra en la construcció i els usos del models metabòlics a escala genòmica per a obtenir eficientment biocombustibles, com etanol i hidrogen. Com a organisme diana, s¿elegí el cianobacteri Synechocystis sp. PCC6803. Aquest organisme ha segut estudiat com una plataforma de producció nodrida per fotons, per la seva habilitat per créixer a partir únicament de diòxid de carboni, aigua i fotons. Aquesta tesi versa sobre mètodes per a modelitzar, analitzar, estimar i predir el comportament metabòlic de cèl¿lules. La principal meta és extreure coneixement del diferents aspectes biològics d¿un organisme de manera que s¿usen per a un objectiu industrial rellevant. La tesi ha segut estructurada en capítols organitzats d¿acord a les successives tasques que acaben construint una cèl¿lula in silico que es comporta, idealment, com la que està basada en carboni. Aquest procés generalment comença amb els arxius de l¿anotació del genoma i acaba amb un model metabòlic a escala genòmica capaç d¿integrar dades ¿òmiques. El primer objectiu de la present tesi és la reconstrucció d¿un model del metabolisme d¿aquest cianobacteri que tinga en compte totes les reaccions que hi estan presents. Esta reconstrucció havia de ser prou flexible com per permetre la simulació del creixement en les diferents condicions ambientals en les quals aquest cianobacteri creix en la natura, així com permetre la integració de diferents nivells d¿informació biològica. Una vegada que aquest requisit fou assolit, es pogueren simular variacions ambientals i estudiar els seus efectes amb una perspectiva de sistema. S¿han simulat fins a cinc condicions de creixement en este model metabòlic i les seves diferències han segut avaluades. La següent tasca fou definir estratègies de producció per a valorar la viabilitat d¿aquest organisme com a plataforma de producció. Es simularen pertorbacions genètiques per al disseny de soques amb producció millorada de metabòlits de rellevància industrial: succinat, etanol i hidrogen. Així, es proposen conjunts de modificacions genètiques per a la sobreproducció d¿aquests metabòlits. També s'han estudiat reaccions acoblades funcionalment i s¿ha ponderat la seva importància en la producció de metabòlits. Finalment, els models metabòlics a escala genòmica permeten establir criteris per integrar diferents tipus de dades que ens ajuden a trobar punts importants de regulació. Eixos centres reguladors, que poden ser objecte de modificacions genètiques, han segut investigats baix canvis dràstics d¿il¿luminació i s¿han inferit principis operacionals del metabolisme. Al llarg d'aquest procés, s¿han revelat punts cecs al metabolisme de Synechocystis sp. PCC6803 i, el més important, punts cecs en la nostra comprensió d'aquest metabolisme. En general, el treball presentat en aquesta tesi dona a conèixer les capacitats industrials del cianobacteri Synechocystis sp. PCC6803 per a produir metabòlits d'interès, tot sent una plataforma de producció neta i sostenible. / Montagud Aquino, A. (2012). Modelling and analysis of biological systems to obtain biofuels [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/17319

[pt] A identificação e categorização de proteínas homólogas são tarefas fundamentais no campo da biologia, que dependem de ferramentas que analisam sequências de nucleotídeos ou aminoácidos. No entanto, a detecção automatizada de padrões evolutivos, assim como outras características, usando métodos tradicionais, ainda apresenta desafios científicos. Neste estudo, propomos uma nova abordagem de representação de dados em camadas, que permite explorar padrões evolutivos e outras características de sequências na busca por similaridades, classificação e agrupamento. Utiliza-se um processo livre de alinhamento e são propostos novos algoritmos de similaridade que permitem aprimorar a eficácia dessa abordagem. Esses algoritmos utilizam técnicas inspiradas na percepção humana para capturar similaridades dentro das representações de moléculas biológicas. Avaliações experimentais demonstram bom desempenho e alta precisão em comparação com abordagens propostas anteriormente. Essa representação em camadas se mostra promissora na identificação de proteínas similares, principalmente com características de homólogas distantes. Além disso, sugere-se também o desenvolvimento de novos métodos e algoritmos de aprendizado de máquina em bioinformática que envolvam a privacidade e segurança de dados biológicos. / [en] The identification and categorization of homologous proteins are fundamental tasks in the field of biology, relying on tools that analyze nucleotide oramino acid sequences. However, automated detection of evolutionary patternsand additional attributes using traditional methods still presents research challenges. In this study, we propose a novel layered data representation approachthat allows us to explore evolutionary patterns and other sequence features insimilarity searching, classification, and clustering. It employs an alignment-freeprocess, and we introduce new similarity algorithms to enhance the effectiveness of this approach. These algorithms leverage techniques inspired by humanperception to capture subtle similarities within biological molecules representations. Experimental evaluations demonstrate good performance and high accuracy compared to previously proposed approaches. This layered representationshows promise in identifying similar proteins, especially with distant homologscharacteristics. Furthermore, it also suggests the development of new methods and machine learning (ML) algorithms in bioinformatics that address theprivacy and security of biological data.

Point datasets that relate to highly populated places, such as ones retrieved from social media or volunteered geographic information in general, can often result in dense point clusters when presented on maps. Therefore, it can be useful to visualize the relevant point density information directly on the urban geometry to tackle the problem of point counting and density range identification in highly cluttered areas. One solution is to relate each point to the nearest geometry object. While this is a straightforward approach, its major drawback is that local point clusters could disappear by assigning them to larger objects, e.g., long roads. To address this issue, we introduce two new point density visualization approaches by which points are related to the underlying geometry objects. In this process, we use grid cells and heatmap contour lines to divide roads, squares, and pedestrian zones into subgeometry units. Comparison of our visualization approaches with conventional density visualization methods shows that our approaches provide a more comprehensive insight into the point distribution over space, i.e., over existing urban geometry. / Wenn Punktdatensätze, die sich auf dicht bevölkerte Räume beziehen – beispielsweise räumliche Daten aus Sozialen Medien oder von VGI-Plattformen – auf Karten dargestellt werden, kommt es häufig zu dichten Punktclustern, was Aussagen über die Anzahl der Punkte oder die Intensität der Punktdichte an bestimmten Orten schwierig bis unmöglich macht. Daher kann es nützlich sein, relevante Informationen über die Punktdichte direkt mit Bezug zu urbanen Geometrien zu visualisieren. Eine Lösung besteht darin, jeden Punkt dem nächstgelegenen Geometrieobjekt zuzuordnen. Ein großer Nachteil dieses Ansatzes ist jedoch, dass lokale Punktcluster verschwinden könnten, indem sie größeren Objekten, z. B. langen Straßen, zugewiesen werden. Um dieses Problem zu lösen, werden zwei neue Ansätze zur Visualisierung der Punktdichte eingeführt, bei denen die Punkte mit den urbanen Geometrieobjekten in Beziehung gesetzt werden, lokale räumliche Eigenschaften jedoch erhalten bleiben. Dafür werden Straßen, Plätze und Fußgängerzonen mithilfe von Rasterzellen und Konturlinien von Kerndichteschätzungen in Teilgeometrieeinheiten unterteilt. Der Vergleich dieser Visualisierungsansätze mit herkömmlichen Dichtevisualisierungsmethoden zeigt, dass die vorgestellten Ansätze einen detaillierteren Einblick in die räumliche Punktverteilung mit Bezug zur bestehenden urbanen Geometrie liefern können.

In the popular sport of football, the exploration of key performance indicators has garnered significant interest among researchers, coaches, and analysts. While machine learning approaches, such as the expected goals model, have provided valuable insights into the attacking aspects of the game, the defensive side has received comparatively less attention. This thesis focuses on the defensive aspect of football, particularly during the opposition’s build-up phase, a strategy increasingly adopted by many teams. The goal of this project is to integrate valuable features from existing research with newly generated ones, developed in consultation with football experts, to create a model that provides insights into a team’s defensive behavior during the opponent’s build-up phase. The study utilizes synchronized event and tracking data from the Allsvenskan 2022 and 2023 seasons. An algorithm is developed to filter and analyze build-up sequences by generating appropriate defensive features. Subsequently, a logistic regression-based machine learning model is implemented to predict the outcome of an event during a build-up sequence, as well as the overall outcome of the sequence. This approach, enables the introduction of two new metrics aimed at evaluating a team’s defensive behavior during the opponent’s build-up phase. Additionally, a web-based application is developed to visualize and communicate the project results and insights to football experts and data analysts. Finally, the findings of this thesis highlight the benefits of combining tracking data with event data in football analytics. / Nyckeltalsundersökningar för att utvärdera och utveckla fotbollsklubbars prestation har väckt stort intresse bland forskare, tränare och analytiker. Traditionellt sett har dessa nyckeltal härletts genom observationsanalys. Dock har den ökande förekomsten av teknik inom den professionella fotbollsvärlden skapat möjligheter för att implementera mer automatiserade metoder för taktisk analys. Denna studie kommer specifikt att fördjupa sig inom det taktiska området av fotboll, som framstår som den mest relevanta och dynamiska delen av spelet. Målet med studien är att applicera modern teknologi, så som maskininlärning, på befintlig forskning, för att utveckla en modell som ger insikter om försvarsspelet under motståndarens uppbyggnadsfas. Studien undersöker synkroniserad händelse- och spårningsdata från Allsvenskan, Sveriges Högsta fotbollsserie, säsongerna 2022 och 2023. Efter att ha utvecklat en algoritm för att filtrera uppbyggnadssekvenser, användes denna data för att generera omfattande egenskaper som beskriver kvaliteten på ett lags försvarsspel. Därefter implementeras en maskininlärningsmodell, med hjälp av logistisk regressionsanalys, för att förutse utfallet av både en uppbyggnadssekvens och försvarsspelet. Resultatet visar på värdet av att kombinera spårningsdata med händelsedata inom fotbollsanalys. Modellens prestanda förbättrades avsevärt, både när det gäller att förutsäga utfallet av ett defensivt spel och en uppbyggnadssekvens.Dessutom har resultatet av studien lett till användbara insikter om försvarsspel för dataanalytiker inom fotboll. En webbaserad applikation utvecklades också för att visualisera och kommunicera resultaten.

Работа посвящена осуществлению комплексного аксиологического и сравнительного анализа русской, английской и китайской видеорекламы автомобильных брендов. Объектом исследования является коммерческая видеореклама автомобилей на русском, английском и китайском языках, предметом – аксиологические, лексико-стилистические и грамматические особенности поликодовых текстов на русском, английском и китайском языках в сфере автомобильной индустрии. Цель исследования заключается в выявлении языковых средств и приемов, которые используются для передачи аксиологических особенностей в видеорекламах автомобильных брендов на русском, английском и китайском языках. Материал исследования составили 513 видеореклам со слоганами (173 на русском языке, 187 на английском, 153 на китайском).Анализ показал, что лексико-стилистические, аксиологические и грамматические особенности видеореклам автомобильных брендов на трех языках имеют значительные отличия, так как каждый язык обладает своими уникальными лексическими единицами, ценностями и грамматическими структурами. Поэтому автомобильным брендам стоит адаптировать слоганы и видеоролики под каждую из целевых аудиторий. Для этого компаниям важно учитывать не только национально-культурные особенности, но и использовать привычные для потенциальных покупателей языковые средства и структуры предложений, а также выстраивать коммуникацию на родном для потребителей языке. Для успешности и эффективности рекламного сообщения автомобильным брендам необходимо учитывать ценности целевой аудитории, а также использовать в слоганах привычные для потребителей языковые формы и структуры. / The study is focused on the implementation of a comprehensive axiological and comparative analysis of the Russian, English and Chinese video advertising of car brands. The object of the study is commercial video advertising of cars in Russian, English and Chinese languages, the subject is axiological, lexico-stylistic and grammatical features of polycoded texts in the Russian, English and Chinese languages in the automotive industry. The aim of the study is to identify the linguistic means and techniques used to convey axiological features in video advertising of car brands in Russian, English and Chinese. The research material includes 513 videos with slogans (173 in Russian, 187 in English, 153 in Chinese).The analysis showed that lexico-stylistic, axiological and grammatical features of car brand video advertising in three languages have significant differences, as each language has its own unique lexical units, values and grammatical structures. Therefore, automotive brands should tailor slogans and videos to each of their target audiences. For this purpose, it is important for companies to consider not only national and cultural specifics, but also to use language tools and sentence structures that are familiar to the potential customers. Thus, for advertising messages to be successful and effective, car brands need to consider the values of the audience and use language forms and structures familiar to the consumers in their slogans.

Η αυτόματη επεξεργασία εικόνων του ενδομητρίου αποτελεί ένα δύσκολο και πολυδιάστατο πρόβλημα, το οποίο έχει απασχολήσει πλήθος ερευνητών και για το οποίο έχει αναπτυχθεί μεγάλος αριθμός τεχνικών. Στην παρούσα διατριβή, παρουσιάζεται μια μεθοδολογική προσέγγιση, η οποία βασίζεται στη χρήση αλγορίθμων ψηφιακής επεξεργασίας και ανάλυσης εικόνων, για την αυτόματη εκτίμηση χαρακτηριστικών που περιγράφουν την αγγείωση και την υφή εικόνων του ενδομητρίου. Αφορμή της μελέτης αποτελεί ο ρόλος που διαπιστώνεται ότι διαδραματίζει η μεταβολή των τιμών των εν λόγω χαρακτηριστικών στην έγκαιρη διάγνωση των παθήσεων του ενδομητρίου. Στα πλαίσια της διατριβής, υλοποιήθηκε κατάλληλη μεθοδολογία για τον υπολογισμό ενός συνόλου χαρακτηριστικών τόσο για υστεροσκοπικές εικόνες, όσο και για ιστολογικές εικόνες του ενδομητρίου. Ιδιαίτερη βαρύτητα δόθηκε στην προ – επεξεργασία των εικόνων προκειμένου να προκύψει βελτίωση της ποιότητας καθώς και ενίσχυση της αντίθεσης αυτών. Στη συνέχεια, ανιχνεύτηκαν τα σημεία που αποτελούν τους κεντρικούς άξονες των υπό εξέταση αγγείων με χρήση διαφορικού λογισμού για τις υστεροσκοπικές εικόνες και υπολογίστηκε ένα σύνολο χαρακτηριστικών μεγεθών που περιγράφουν την αγγείωση και την υφή των εικόνων τόσο για τις υστεροσκοπικές όσο και για τις ιστολογικές εικόνες. Τέλος, εφαρμόστηκαν κατάλληλοι αλγόριθμοι με σκοπό την κατηγοριοποίηση των υστεροσκοπικών και των ιστολογικών εικόνων και συγκεκριμένα τον διαχωρισμό των παθολογικών και των φυσιολογικών εικόνων του ενδομητρίου. Παράλληλα, χρησιμοποιήθηκε η ROC ανάλυση στην απεικόνιση και ανάλυση της συμπεριφοράς των εν λόγω κατηγοριοποιητών. / Automatic analysis of the endometrial images is a difficult and multidimensional problem. For this reason, the number of papers and techniques regarding this issue is numerous. In this Thesis, a methodology is presented, based on advance image processing techniques in order to automatically estimate texture and vessel’s features in endometrial images. Motivation for the Thesis is the fact that the variation of the measurements of the specific features plays significant role in the seasonable diagnosis of endometrial disorders. Throughout this Thesis, an appropriate methodology is developed in order to estimate the features for the hysteroscopical and histological images of the endometrium. An important step is the pre – processing of the images in order to enhance the image quality and the image contrast. Then, the pixels that constitute the centerlines of vessels are detected by using differential calculus for the hysteroscopical images, only. Furthermore, the texture and vessel’s features in hysteroscopical and histological images are estimated. Finally, appropriate algorithms are applied in order to classify the hysteroscopical and histological images and distinguish pathological and normal endometrial images. ROC analysis is used in order to evaluate the discrimination power of the features that were estimated.

Visual Simultaneous Localisation and Mapping (SLAM) is concerned with simultaneously estimating the pose of a camera and a map of the environment from a sequence of images. Traditionally, sparse maps comprising isolated point features have been employed, which facilitate robust localisation but are not well suited to advanced applications. In this thesis, we present map representations that allow a more dense description of the environment. In one approach, planar features are used to represent textured planar surfaces in the scene. This model is applied within a visual SLAM framework based on the Extended Kalman Filter. We presents solutions to several challenges which arise from this approach.

This thesis deals with content based automatic photo categorization. The aim of the work is to experiment with advanced techniques of image represenatation and to create a classifier which is able to process large image dataset with sufficient accuracy and computation speed. A traditional solution based on using visual codebooks is enhanced by computing color features, soft assignment of visual words to extracted feature vectors, usage of image segmentation in process of visual codebook creation and dividing picture into cells. These cells are processed separately. Linear SVM classifier with explicit data embeding is used for its efficiency. Finally, results of experiments with above mentioned techniques of the image categorization are discussed.