Search

This paper presents KD SENSO-MERGER, a novel Knowledge Discovery (KD) architecture that is capable of semantically integrating heterogeneous data from various sources of structured and unstructured data (i.e. geolocations, demographic, socio-economic, user reviews, and comments). This goal drives the main design approach of the architecture. It works by building internal representations that adapt and merge knowledge across multiple domains, ensuring that the knowledge base is continuously updated. To deal with the challenge of integrating heterogeneous data, this proposal puts forward the corresponding solutions: (i) knowledge extraction, addressed via a plugin-based architecture of knowledge sensors; (ii) data integrity, tackled by an architecture designed to deal with uncertain or noisy information; (iii) scalability, this is also supported by the plugin-based architecture as only relevant knowledge to the scenario is integrated by switching-off non-relevant sensors. Also, we minimize the expert knowledge required, which may pose a bottleneck when integrating a fast-paced stream of new sources. As proof of concept, we developed a case study that deploys the architecture to integrate population census and economic data, municipal cartography, and Google Reviews to analyze the socio-economic contexts of educational institutions. The knowledge discovered enables us to answer questions that are not possible through individual sources. Thus, companies or public entities can discover patterns of behavior or relationships that would otherwise not be visible and this would allow extracting valuable information for the decision-making process.

The objective of Geo-IA is to research, design and implement a Geo-Smart Artificial Intelligence technology platform for public and private business organizations. The GeoIA project presents a geolocation platform that integrates technological innovation to support a strategy for the creation of a Smart Territories. To do this, Text Mining, Machine Learning (including deep learning) and Natural Language Processing technologies are deployed. The functionality of the geolocation platform is to analyze, integrate, share data, visualize and represent territorial indicators, with the aim of facilitating the monitoring and fulfillment of territorial strategies. In short, GeoIA promotes interoperability between public administration bodies and also provides citizens with mechanisms to access information of interest, where the magnitude of the integrated and interrelated data permits. GeoIA also provides digital knowledge (tools, linked information, semantics, virtual assistants) for use by public administrations to enhance their decision making through greater knowledge of the environment and to improve services to citizens.

This work emphasizes the computational and analytical analysis of integral-differential equations, with a particular application in modeling avoidance learning processes. Firstly, we suggest an approach to determine a unique solution to the given model by employing methods from functional analysis and fixed-point theory. We obtain numerical solutions using the approach of Picard iteration and evaluate their stability in the context of minor perturbations. In addition, we explore the practical application of these techniques by providing two examples that highlight the thorough analysis of behavioral responses using numerical approximations. In the end, we examine the efficacy of our suggested ordinary differential equations (ODEs) for studying the avoidance learning behavior of animals. Furthermore, we investigate the convergence and error analysis of the proposed ODEs using multiple numerical techniques. This integration of theoretical and practical analysis enhances the domain of applied mathematics by providing important insights for behavioral science research.

This study undertakes a comprehensive analysis of second-order Ordinary Differential Equations (ODEs) to examine animal avoidance behaviors, specifically emphasizing analytical and computational aspects. By using the Picard–Lindelöf and fixed-point theorems, we prove the existence of unique solutions and examine their stability according to the Ulam-Hyers criterion. We also investigate the effect of external forces and the system’s sensitivity to initial conditions. This investigation applies Euler and Runge–Kutta fourth-order (RK4) methods to a mass-spring-damper system for numerical approximation. A detailed analysis of the numerical approaches, including a rigorous evaluation of both absolute and relative errors, demonstrates the efficacy of these techniques compared to the exact solutions. This robust examination enhances the theoretical foundations and practical use of such ODEs in understanding complex behavioral patterns, showcasing the connection between theoretical understanding and real-world applications.

This study investigates the intricacies of animal decision-making in T-maze environments through a synergistic approach combining computational modeling and machine learning techniques. Focusing on the binary decision-making process in T-mazes, we examine how animals navigate choices between two paths. Our research employs a mathematical model tailored to the decision-making behavior of fish, offering analytical insights into their complex behavioral patterns. To complement this, we apply advanced machine learning algorithms, specifically Support Vector Machines (SVM), K-Nearest Neighbors (KNN), and a hybrid approach involving Principal Component Analysis (PCA) for dimensionality reduction followed by SVM for classification to analyze behavioral data from zebrafish and rats. The above techniques result in high predictive accuracies, approximately 98.07% for zebrafish and 98.15% for rats, underscoring the efficacy of computational methods in decoding animal behavior in controlled experiments. This study not only deepens our understanding of animal cognitive processes but also showcases the pivotal role of computational modeling and machine learning in elucidating the dynamics of behavioral science.

Large language models have shown impressive performance in Natural Language Processing tasks, but their black box characteristics render the explainability of the model’s decision difficult to achieve and the integration of semantic knowledge. There has been a growing interest in combining external knowledge sources with language models to address these drawbacks. This paper, OntoLM, proposes a novel architecture combining an ontology with a pre-trained language model to classify biomedical entities in text. This approach involves constructing and processing graphs from ontologies and then using a graph neural network to contextualize each entity. Next, the language model and the graph neural network output are combined into a final classifier. Results show that OntoLM improves the classification of entities in medical texts using a set of categories obtained from the Unified Medical Language System. We can create more traceable natural language processing architectures using ontology graphs and graph neural networks., Los grandes modelos de lenguaje han mostrado un rendimiento impresionante en tareas de Procesamiento del Lenguaje Natural, pero su condición de caja negra hace difícil explicar las decisiones del modelo e integrar conocimiento semántico. Existe un interés creciente en combinar fuentes de conocimiento externas con LLMs para solventar estos inconvenientes. En este artículo, proponemos OntoLM, una arquitectura novedosa que combina una ontología con un modelo de lenguaje pre-entrenado para clasificar entidades biomédicas en texto. El enfoque propuesto consiste en construir y procesar grafos provenientes de una ontología utilizando una red neuronal de grafos para contextualizar cada entidad. A continuación, combinamos los resultados del modelo de lenguaje y la red neuronal de grafos en un clasificador final. Los resultados muestran que OntoLM mejora la clasificación de entidades en textos médicos utilizando un conjunto de categorías obtenidas de Unified Medical Language System. Utilizando grafos de ontologías y redes neuronales de grafos podemos crear arquitecturas de procesamiento de lenguaje natural más rastreables.

La enorme cantidad de información disponible en la web, en múltiples formatos, requiere el diseño de sistemas de software para aprovechar esta información y obtener conocimiento válido de ella. En este contexto, surgen interesantes retos en el descubrimiento de conocimiento relevante en diversas fuentes de datos no estructurados (como blogs, sitios web y medios sociales) y su integración con el conocimiento presente en fuentes estructuradas (como bases de datos o redes semánticas). Además, la velocidad a la que se produce la nueva información crea nuevos retos, relacionados con el procesamiento continuo, la evaluación y el refinamiento del conocimiento adquirido. Este trabajo consiste en proponer y desarrollar una arquitectura para el descubrimiento continuo e incremental de conocimiento en múltiples fuentes de texto estructurado y no estructurado.

The large amount of information in digital format that exists today makes it unfeasible to use manual means to acquire the knowledge contained in these documents. Therefore, it is necessary to develop tools that allow us to incorporate this knowledge into a structure that is easy to use by both machines and humans. This paper presents a system that can incorporate the relevant information from a document in any format, structured or unstructured, into a semantic network that represents the existing knowledge in the document. The system independently processes from structured documents based on its annotation scheme to unstructured documents, written in natural language, for which it uses a set of sensors that identifies the relevant information and subsequently incorporates it to enrich the semantic network that is created by linking all the information based on the knowledge discovered.

The project T2Know presents the use of natural language processing technologies for the creation of a semantic platform of scientific documents via knowledge graphs. This knowledge graph will link relevant parts of each document with those of other documents in such a way that trend analysis and recommendations can be achieved. The goals addressed within the scope of this project include entity recognizers development, profile definition and documents linkage through the use of transformers technologies. As a result, the relevant parts of the documents to be extracted are related not only to the title and affiliation of the authors, but also to article topics such as references, which are also considered relevant parts of the scientific article.

Generative language models achieve the state of the art in many tasks within natural language processing (NLP). Although these models correctly capture syntactic information, they fail to interpret knowledge (semantics). Moreover, the lack of interpretability of these models promotes the use of other technologies as a replacement or complement to generative language models. This is the case with research focused on incorporating knowledge by resorting to knowledge bases mainly in the form of graphs. The generation of large knowledge graphs is carried out with unsupervised or semi-supervised techniques, which promotes the validation of this knowledge with the same type of techniques due to the size of the generated databases. In this review, we will explain the different techniques used to test and infer knowledge from graph structures with machine learning algorithms. The motivation of validating and inferring knowledge is to use correct knowledge in subsequent tasks with improved embeddings.

A la hora de desarrollar una aplicación, es muy importante plantear una buena arquitectura para así poder tener un buen punto de partida. Las aplicaciones modernas suelen tener una arquitectura muy bien definida, lo que se traduce en un bajo acoplamiento, el uso de microservicios o APIs para la interconexión de datos, o la definición de diferentes capas entre las cuales se reparte el código. Las ontologías son una de las formas de representación de conocimiento en formato digital más utilizadas. Learning Engine Through Ontologies (LETO) es una aplicación desarrollada conjuntamente por el Grupo de Procesamiento de Lenguaje y Sistemas de Información de la Universidad de Alicante (GPLSI) y la Escuela de Matemáticas y Computación de la Universidad de la Habana (Cuba). El objetivo de la aplicación es permitir introducir información de diferentes fuentes, tanto estructuradas como no estructuradas, y traducir esa información a ontologías. Una vez se tienen esas ontologías almacenadas, se contará con una base de conocimiento la cual podrá ser representada utilizando diferentes visualizadores (mapas, histogramas, gráficos…). Esa representación vendrá dada por una consulta realizada por el usuario en lenguaje natural, la cual será procesada para extraer de la base de datos la información relevante con la consulta realizada. LETO actualmente cuenta con diversos problemas derivados de la falta de una arquitectura clara. Durante el presente trabajo se plantea una arquitectura de tres capas para la aplicación, lo que permitirá desacoplar el código existente al crear una nueva capa de presentación desde cero, así como una API para realizar el intercambio de datos. La aplicación implementa una serie de sensores que utilizan modelos de inteligencia artificial para procesar los textos introducidos. Esos sensores también se encontraban acoplados, por lo que se plantea una nueva estructura para que sea más sencilla la gestión e implementación de sensores. Para poder realizar esto, se h