Artículos y presentaciones en Congresos

URI permanente para esta colecciónhttps://digital.cic.gba.gob.ar/handle/11746/543

Examinar

Envíos recientes

Mostrando 1 - 20 de 310
  • Documento de conferencia
    Acceso Abierto
    Teaching digital humanities at high school level
    (2025) Lliteras, Alejandra Beatriz; Artopoulos, Alejandro Martín
    The incorporation of digital technology in educational environments was usually thought of from a didactic and/or pedagogical perspective. However, the digital transformation driven by smartphones, platforms and generative artificial intelligence showed that curricular and/or epistemological aspects had to be considered, such as the incorporation of programming and computational thinking in basic education. Within this line of research, it is particularly productive to inquire about the incorporation of digital humanities at the high school level. The objective of this work is to know through a first literature review, the implementation of educational experiences in the field of digital humanities at the high school level of articles from academic sources. In addition to present global and local initiatives of innovative aspects in digital education. The analysis addresses the description of the fields in which relevant cases of teaching experiences were found, the digital methods used as well as the emergence of new emerging curricular areas such as Digital Citizenship. The conclusions point out the areas of opportunity for the development of knowledge areas, topics, applications and didactic materials to continue the development of the teaching of humanistic or digital humanistic computational knowledge at the high school level.
  • Documento de conferencia
    Acceso Abierto
    Adversarial image generation using geneticalgorithms with black-box technique
    (2023) Pérez, Gabriela Alejandra; Pons, Claudia Fabiana
    Convolutional neural networks are a technique that has demon-strated great success in computer vision tasks, such as image classifica-tion and object detection. Like any machine learning model, they havelimitations and vulnerabilities that must be carefully considered for safeand effective use. One of the main limitations lies in their complexityand the difficulty of interpreting their internal workings, which can beexploited for malicious purposes. The goal of these attacks is to makedeliberate changes to the input data in order to deceive the model andcause it to make incorrect decisions. These attacks are known as adver-sarial attacks. This work focuses on the generation of adversarial im-ages using genetic algorithms for a convolutional neural network trainedon the MNIST dataset. Several strategies are employed, including tar-geted and untargeted attacks, as well as the presentation of interpretableand non-interpretable images that are unrecognizable to humans but aremisidentified and confidently classified by the network. The experimentdemonstrates the ability to generate adversarial images in a relativelyshort time, highlighting the vulnerability of neural networks and the easewith which they can be deceived. These results underscore the impor-tance of developing more secure and reliable artificial intelligence systemscapable of resisting such attacks
  • Artículo
    Acceso Abierto
    Semi-supervised learning models for documentclassification: A systematic review and meta-analysis
    (2023) Cevallos-Culqui, Alex; Pons, Claudia Fabiana; Rodríguez, Gustavo
    The proliferation of digital documents in the internet has given rise to the search for informationpatterns that allow for the categorization of organizational documents to generate knowledge in a institution.One of the Artificial Intelligence techniques for this purpose is text classification, which for its application useslabels (categorized documents) with supervised (with labels) or unsupervised (without labels) training models.Both traditional models with their advantages and disadvantages have been consolidated into semi-supervisedmodels to extract the best qualities of each one, however, the labeling process involves resources that need to beoptimized to improve the classification accuracy. An analysis of the types of semi-supervised models would showthe strengths of their training and how the structure of each of them affects the accuracy of their classification. Thepresent study proposes a structure of semi-supervised model in document classification, in order to analyze thequalities of each one in their categorization process, it through a systematic literature review (SLR) that analyzesthe performance of the studies to conduct a meta-analysis. Further, the study search strategy was defined by thePICOC method (Population, Intervention, Comparison, Outcome, Context), supported by two research questionsand delimited in a search chain that allowed the collection of 332 research studies. These papers were filteredusing the PRISMA method and the determination of exclusion criteria, in total 46 papers have been selected forthe present study.From this SLR, an organizational structure has been obtained for semi-supervised models anda scheme for the classification process. In addition, the advantages and disadvantages of different learning typeshave been analyzed, evaluating their classification performance in each type of learning through a meta-analysis.This has determined that the models that present the best levels of performance are active learning model (0.88)and ensemble learning model (0.84)
  • Artículo
    Acceso Abierto
    Rule Extraction in Trained Feedforward Deep Neural Networks: Integrating Cosine Similarity and Logic for Explainability
    (2024) Negro, Pablo Ariel; Pons, Claudia Fabiana
    Explainability is a key aspect of machine learning, necessary for ensuring transparency and trust in decision-making processes. As machine learning models become more complex, the integration of neural and symbolic approaches has emerged as a promising solution to the explainability problem. One effective solution involves using search techniques to extract rules from trained deep neural networks by examining weight and bias values and calculating their correlation with outputs. This article proposes incorporating cosine similarity in this process to narrow down the search space and identify the critical path connecting inputs to final results. Additionally, the integration of first-order logic (FOL) is suggested to provide a more comprehensive and interpretable understanding of the decision-making process. By leveraging cosine similarity and FOL, an innovative algorithm capable of extracting and explaining rule patterns learned by a feedforward trained neural network was developed and tested in two use cases, demonstrating its effectiveness in providing insights into model behavior.
  • Documento de conferencia
    Acceso Abierto
    Extracción de reglas de redes neuronales feedforward entrenadas con lógica de primer orden
    (2024) Negro, Pablo; Pons, Claudia Fabiana
    La necesidad de integración neural-simbólica se hace evidente a me-dida que se abordan problemas más complejos, y que van más allá de tareas de dominio limitadas como lo es la clasificación. Los métodos de búsqueda para la extracción de reglas de las redes neuronales funcionan enviando combinaciones de datos de entrada que activan un conjunto de neuronas. Ordenando adecuada-mente los pesos de entrada de una neurona, es posible acotar el espacio de bús-queda. Con base en esta observación, este trabajo tiene por objetivo presentar un método para extraer el patrón de reglas aprendido por una red neuronal entrenada feedforward, analizar sus propiedades y explicar estos patrones a través del uso de lógica de primer orden (FOL)
  • Artículo
    Acceso Abierto
    Evaluation of Transfer Learning Techniques in Neural Networks with Tiny-scale Training Data
    (2023) Pérez, Gabriela Alejandra; Jacinto, Milagros; Moschettoni, Martín; Pons, Claudia Fabiana
    This paper rigorously analyzes the process of building a deep neural network for image recognition and classification using Transfer Learning techniques. The biggest challenge is assuming that the training dataset is very small. The research is based on addressing a particular case study, the income of donations to the Food Bank of La Plata. The results obtained corroborate that the techniques analyzed are appropriate to solve tasks of detection and classification of images even in cases in which there is a very moderate number of samples.
  • Documento de conferencia
    Acceso Abierto
    A baseline underwater soundscape of an intensely human-exploited estuarine and the effects of vessel traffic sound
    (2024) Pons, Juan; Uibrig, Román; Molina, Juan; Pons, Claudia Fabiana
    In this article, we studied the anthropically impacted natural environmental sound in the port of Bahía Blanca, located in the southern province of Buenos Aires, Argentina. To acquire the acoustic signals, an omni-directional passive hydrophone was used. The acoustic signals were analyzed using scripts implemented in the R programming language. Temporal series without maritime traffic were used as a baseline to describe the soundscape in the harbour area by estimating its power spectral density (PSD). Subsequently, the acoustic environment was analyzed with the presence of two man-made acoustic sources: "boat" and "ship" in the vicinity. Finally, the calculated normal soundscape level in the harbour has a magnitude of 116.25 dB re 1 µPa.
  • Documento de conferencia
    Acceso Abierto
    Método_SCGE validando las cuatro etapas en un organismo gubernamental
    (2024) Castro, Marcelo; Pons, Claudia Fabiana; Rodríguez, Rocío Andrea
    En el presente trabajo se realiza la validación de las cuatro etapas correspondientes a la metodología para sistematizar y estandarizar los procesos de Gobierno Electrónico en la gestión pública, a través de servicios computacionales (metodo_SCGE). Esta metodología considera software, hardware y comunicaciones, el análisis de estos componentes permite generar un modelo integral basado en Servicios Computacionales para Gobierno Electrónico (SCGE). El artículo contiene una presentación de algunas metodologías existentes que se pueden aplicar a servicios computacionales, una breve descripción de metodo_SCGE, incluyendo componentes, características, etapas y actividades. Al final se presenta la validación de las cuatro etapas, sobre un servicio específico perteneciente a un organismo gubernamental.
  • Documento de conferencia
    Acceso Abierto
    Identification of biological properties in organismsusing Machine Learning techniques on wholegenome sequences
    (2023) Ferella, Nicolas; Pizio, Pablo
    The advance in technology and genome sequencing processes in the recentdecades have made large volumes of biological data available to researchers fromall over the world, which, due to the large scales, are difficult to analyze in theirentirety. Therefore, it is intuitive to think of Artificial Intelligence to work withsuch information.In order to reduce the existing gap between the researchers and the ArtificialIntelligence tools, a software was developed that allows the creation of a works-pace for biological organisms, the processing of its corresponding genomes, andthe creation and training of models of Machine Learning, everything using asimple (yet powerful) graphical interface.The trained models are then analyzed to find which patterns determine theresult of the property that is being investigated on the biological organism,finding in the process the genes with the greatest impact on the model’s predic-tions, allowing the researcher to subsequently analyze the desired genes in thelaboratory, saving time and resources in the process
  • Documento de conferencia
    Acceso Abierto
    El desafío de Scrum distribuido en diferentes locaciones
    (2023) Salazar, Joaquín; Grimaldi, Pablo; Pons, Claudia
    En las últimas décadas la tecnología ha avanzado rápidamente y con ella la forma de trabajo de todas las personas relacionadas con IT, hoy en día es totalmente normal que un equipo esté integrado por personas que están en diferentes ciudades del mundo, trabajando de manera remota o con diferentes husos horarios e idiomas. Al mismo tiempo, el uso de las metodologías ágiles; principalmente Scrum, ha tenido un gran crecimiento en su implementación. Por esta razón es oportuno poder realizar un análisis de todos los desafíos que implica usar Scrum de manera distribuida, brindando además un aporte de posibles soluciones y consejos para afrontarlos.
  • Documento de conferencia
    Acceso Abierto
    Validando la segunda etapa de metodo_SCGE en un organismo gubernamental
    (2023) Castro, Marcelo; Pons, Claudia Fabiana; Rodríguez, Rocío Andrea
    En el presente trabajo se realiza la validación de la segundaetapa correspondiente a la metodología para sistematizar y estandarizar los procesos de Gobierno Electrónico en la gestión pública, a través de servicios computacionales (metodo_SCGE). En esta metodología se consideran distintos aspectos tales como software, hardware y comunicaciones, el análisis de estos componentes permite generar un modelo integral basado en Servicios Computacionales de Gobierno Electrónico (SCGE). El artículo contieneuna presentación de algunas metodologías existentes que se pueden aplicar a servicios computacionales, una breve introducción al concepto de servicios computacionales en el ámbito de gobierno electrónico y una descripción de sus componentes. También se describen las etapas que componen la metodología desarrollada. Finalmente se presenta la validaciónde la segunda etapasobre un servicio específico perteneciente a un organismo gubernamental
  • Artículo
    Acceso Abierto
    Semiotics: An Approach to Model Security Scenarios for IoT-Based Agriculture Software
    (2024) Hurtado, Julio Ariel; Antonelli, Leandro; López, Santiago; Gómez, Adriana; Delle Ville, Juliana; Maltempo, Giuliana; Zambrano, Frey Giovanny; Solis, Andrés; Camacho, Marta Cecilia; Solinas, Miguel; Kaplan, Gladys; Muñoz, Freddy
    Agriculture is a vital human activity that contributes to sustainable development. A few decades ago, the agricultural sector adopted the Internet of Things (IoT), which has played a relevant role in precision and smart farming. The IoT developments in agriculture require that numerous connected devices work cooperatively. This increases the vulnerability of IoT devices, mainly because they lack the necessary built-in security because of their context and computational capacity. Other security threats to these devices are related to data storage and processing connected to edge or cloud servers. To ensure that IoT-based solutions meet functional and non-functional requirements, particularly those concerning security, software companies should adopt a security-focused approach to their software requirements specification. This paper proposes a method for specifying security scenarios, integrating requirements and architecture viewpoints into the context of IoT for agricultural solutions. The method comprises four steps: (i) describe scenarios for the intended software, (ii) describe scenarios with incorrect uses of the system, (iii) translate these scenarios into security scenarios using a set of rules, and (iv) improve the security scenarios. This paper also describes a prototype application that employs the proposed algorithm to strengthen the incorrect use scenario based on the correct use scenario. Then, the expert can complete the information for the analysis and subsequent derivation of the security scenario. In addition, this paper describes a preliminary validation of our approach. The results show that the proposed approach enables software engineers to define and analyze security scenarios in the IoT and agricultural contexts with good results. A survey administered to five security experts found that the proposed security scenario method is generally useful for specifying agricultural IoT solutions but needs improvement in different areas.
  • Documento de conferencia
    Acceso Abierto
    Pensar Ágora como un dispositivo para la ciencia participativa desde el sur global
    (2024) Torres, Diego; Katzer, Leticia
    La ciencia participativa es una práctica de la ciencia abierta en que personas vinculadas y no vinculadas a instituciones científicas colaboran para llevar a cabo un proyecto científico en forma conjunta. En estos proyectos suele diferenciarse a los participantes entre científicos profesionales, y científicos ciudadanos. Los profesionales son aquellas personas que participan del proyecto y poseen una vinculación con una entidad académica incluyendo los alcances del proyecto dentro de sus actividades laborales. Por otro lado, las personas científicas ciudadanas son aquellas que se vinculan al proyecto pero no poseen un vínculo institucional con una entidad relacionada al proyecto y por consiguiente, sus actividades laborales no se vinculan con el proyecto. La ciencia participativa, también conocida como ciencia ciudadana, posee diferentes niveles de participación de personas no profesionales de la ciencia. Dependiendo del vínculo y de la toma de decisiones los proyectos pueden clasificarse en proyectos de contributivos, de colaboración, de cooperación o co-creados. En los primeros los participantes no académicos son relegados solamente a tareas de recolección y las desiciones de diseño los toman los miembros académicos, mientras que en los últimos, todos los procesos y tomas de decisiones del proyecto se definen en conjunto por todos los perfiles de participantes. Los proyectos cocreados son los que dan forma a un proyecto realmente participativo. Ágora es una plataforma que permite la co-creación de proyectos colectivos con especial foco en aquellos que utilizan un Smart Phone como herramienta de recolección, sin necesidad de tener habilidades de programacion. Cualquier colectivo puede co-crear sus proyectos e inmediatamente usarlos. Dado que la tecnología no es neutral, el propósito principal del proyecto Ágora es brindar una herramienta que permita a una comunidad de personas articular sus proyectos de ciencia ciudadana de recolección de una forma sencilla y sin necesidad de desarrollos costosos de aplicaciones personalizadas.
  • Documento de conferencia
    Acceso Abierto
    Diseño de indagación para identificar el uso de TIC en la agricultura familiar
    (2024) Riva, María Florencia; Challiol, Cecilia; Del Pino, Mariana; Fernández, Alejandro
    La agricultura familiar del Cinturón Hortícola del Gran La Plata es un sector de vital relevancia en la economía regional. El uso de las Tecnologías de la Información y Comunicación (TIC) en esta población es todo un interrogante. Esta falta de datos significativos impacta en el diseño y desarrollo de futuras soluciones tecnológicas adecuadas para estos sujetos. Lograr empatizar con las necesidades reales de las personas es esencial para el diseño y desarrollo de software. Esto se vuelve un desafío mayor cuando hay un equipo multidisciplinario involucrado en la tarea de indagación. Esta ponencia tiene como objetivo presentar y discutir los lineamientos para el diseño de una metodología de indagación que combina encuestas semiestructuradas y entrevistas para empatizar, para obtener un conocimiento profundo, tanto cuantitativo como cualitativo. Participaron de la formulación de esta metodología un equipo multidisciplinario, que incluyó a una socióloga, una ingeniera agrónoma y dos informáticos; dado que se busca indagar en relación con el uso de TIC por parte de los/as trabajadores/as del Cinturón Hortícola del Gran La Plata. A lo largo de la ponencia, de carácter exploratorio, se presenta la metodología diseñada, detallando cómo las encuestas proporcionaron una primera aproximación cuantitativa al campo, y cómo las entrevistas permitieron profundizar en la comprensión cualitativa de las necesidades tecnológicas de los sujetos. Se espera que la metodología de indagación presentada, la cual involucra a un equipo multidisciplinario, sirva de guía para que otros equipos de trabajo la puedan extrapolar a otros dominios de interés.
  • Revisión
    Acceso Abierto
    Designing Microservices Using AI: A Systematic Literature Review
    (2025) Narváez, Daniel; Battaglia, Nicolas; Fernández, Alejandro; Rossi, Gustavo Héctor
    Microservices architecture has emerged as a dominant approach for developing scalable and modular software systems, driven by the need for agility and independent deployability. However, designing these architectures poses significant challenges, particularly in service decomposition, inter-service communication, and maintaining data consistency. To address these issues, artificial intelligence (AI) techniques, such as machine learning (ML) and natural language processing (NLP), have been applied with increasing frequency to automate and enhance the design process. This systematic literature review examines the application of AI in microservices design, focusing on AI-driven tools and methods for improving service decomposition, decision-making, and architectural validation. This review analyzes research studies published between 2018 and 2024 that specifically focus on the application of AI techniques in microservices design, identifying key AI methods used, challenges encountered in integrating AI into microservices, and the emerging trends in this research area. The findings reveal that AI has effectively been used to optimize performance, automate design tasks, and mitigate some of the complexities inherent in microservices architectures. However, gaps remain in areas such as distributed transactions and security. The study concludes that while AI offers promising solutions, further empirical research is needed to refine AI’s role in microservices design and address the remaining challenges.
  • Documento de conferencia
    Acceso Abierto
    Automatización de documentos judiciales utilizando Inteligencia Artificial
    (2024) Pérez, Gabriela Alejandra; Picasso, Nicolas; Mostaccio, Catalina Alba; Antonelli, Leandro
    La automatización de documentos (DA) busca reducir la intervención manual en la generación, gestión y procesamiento de documentos. Este enfoque es especialmente beneficioso para documentos altamente estructurados como informes legales, técnicos y clínicos. En este contexto, la integración de tecnologías avanzadas, como el procesamiento de lenguaje natural (NLP) y los grandes modelos de lenguaje (LLM), ofrece soluciones innovadoras para mejorar la eficiencia y precisión en la gestión de documentos. Este trabajo presenta una herramienta diseñada para crear plantillas a partir de primeros despachos, extrayendo información relevante sin incluir datos sensibles. La herramienta permite la personalización de documentos mediante elementos opcionales, con el objetivo de estandarizar y mejorar la eficiencia en la elaboración de textos legales. La contribución principal de este estudio es la integración de LLMs de código abierto, como LLaMA, y la aplicación de técnicas de auto-refinamiento para optimizar la precisión y relevancia del procesamiento textual.
  • Artículo
    Acceso Abierto
    Evaluation of natural language processing models to measure similarity between scenarios written in Spanish
    (2024) Pérez, Gabriela Alejandra; Mostaccio, Catalina Alba; Antonelli, Leandro; Maltempo, Giuliana
    Requirements engineering is a critical phase in software development; it seeks to understand and document system requirements from early stages. Typically, requirements specification involves close collaboration be- tween customers and development teams. Customers contribute their expertise in the domain language, while developers use more technical, computational terms. Despite these differences, achieving mutual understanding is crucial. One of the most widely used artifacts for this purpose is scenarios. In environments where multiple actors write scenarios, duplication is common. Thus, there is a need for mechanisms to detect similar scenarios and prevent redundancy. In this paper we empirically evaluate several pre-trained Natural Language Processing models to analyze the semantic similarity between scenarios in Spanish, identifying words or phrases with equivalent meanings. It is important to note that the analysis is performed in this language to contribute to the region. Finally, we present a tool that facilitates the creation of new scenarios by identifying potential similarities with existing ones. The tool supports multiple models, allowing users to select the most appropriate one to detect similarscenarios accurately during the definition process.
  • Documento de conferencia
    Acceso Abierto
    Integración de diferentes técnicas para visualizar la influencia de regiones de una imagen en su clasificación por una red neuronal
    (2024) Gardella Ruiz, Andrés; Pérez, Gabriela Alejandra; Pons, Claudia Fabiana
    Hoy en día es común la utilización, en múltiples ámbitos, de redes neuronales que permiten realizar actividades complejas como clasificación de imágenes. Si bien es una tecnología muy útil debido a la información que provee, en contextos sensibles como el de la salud pública, es necesario poder entender y confiar en dicha información, ya que la falta de precisión puede acarrear consecuencias negativas significativas. Esta necesidad de comprender el funcionamiento y la toma de decisiones de las redes neuronales, ha dado lugar al surgimiento de métodos y técnicas de visualización, que permiten comprender mejor las decisiones tomadas por éstas en base a la información ingresada. Este trabajo tiene como propósito analizar algunos de estos métodos de visualización para luego desarrollar una herramienta que simplifique su uso y la visualización de las explicaciones. La herramienta permitirá comparar los resultados y facilitará la interpretación de las decisiones de la red, haciendo que estos métodos sean más accesibles.
  • Documento de conferencia
    Acceso Abierto
    Quantum Computing and Its Resources in Software Engineering
    (2023) Lammers, Marcos
    This research aims to identify and characterize quantum computing resources for software engineering, considering NISQ device limitations.
  • Documento de conferencia
    Acceso Abierto
    Clustering Tasks and Decision Trees with Augustan Love Poets: Cohesion and Separation in Feature Importance Extraction
    (2024) Nusch, Carlos Javier; del Rio Riande, Gimena; Cagnina, Leticia Cecilia; Errecalde, Marcelo Luis; Antonelli, Leandro
    This article extends various automatic text analysis tasks from previous works by applying natural language processing techniques to a corpus of Latin texts from the 1st century BC and 1st century AD. The motivation behind this work is to delve into and understand a historical literary trend revolving around the themes of love> spanning from antiquity through to the medieval period. The analyzed authors include Gaius Valerius Catullus’ Albius Tibullus’ and Sextus Propertius’ representing the literary movement of the neoterics’ and Publius Vergilius Maro and Marcus Annaeus Lucanus’ epic poets with distinct styles’ serving as control samples. Unlike previous works’ various corrections were added to the preprocessing tasks’ including improved word tokenization with enclitics and handling of orthographic variances. For the clustering tasks’ the K-Means method and the Silhouette Score were used to determine the optimal cluster sizes. Using these optimal clusters as labels’ decision trees were trained for each range of n-grams’ aiming to identify features with the highest Information Gain and Information Gain Ratio. The trees were trained based on the criterion of Entropy’ and calculations of Feature Importance were performed. In this study’ we focused on detailing the classification results and features extracted by the decision trees’ based on the best Silhouette scores obtained and the Information Gain. We examined whether the words or parts of words with classificatory potential identified in the process matched the findings from previous exploratory tasks performed using other techniques.