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.

Annotated corpora are indispensable tools to train computational models in Natural Language Processing. However, in the case of more complex semantic annotation processes, it is a costly, arduous, and time-consuming task, resulting in a shortage of resources to train Machine Learning and Deep Learning algorithms. In consideration, this work proposes a methodology, based on the human-in-the-loop paradigm, for semi-automatic annotation of complex tasks. This methodology is applied in the construction of a reliability dataset of Spanish news so as to combat disinformation and fake news. We obtain a high quality resource by implementing the proposed methodology for semi-automatic annotation, increasing annotator efficacy and speed, with fewer examples. The methodology consists of three incremental phases and results in the construction of the RUN dataset. The annotation quality of the resource was evaluated through time-reduction (annotation time reduction of almost 64% with respect to the fully manual annotation), annotation quality (measuring consistency of annotation and inter-annotator agreement), and performance by training a model with RUN semi-automatic dataset (Accuracy 95% F1 95%), validating the suitability of the proposal.

El aprendizaje automático es un campo de la inteligencia artificial que ha ganado un reciente interés en todas las áreas de la industria, motivado fundamentalmente por el acelerado crecimiento de las capacidades de cómputo y la disponibilidad de datos. Sin embargo, una de las principales dificultades para su aplicación es la necesidad de expertos que conozcan los detalles internos de los múltiples modelos que pueden ser utilizados. En este contexto ha surgido un nuevo campo de estudio, denominado AutoML (Automated Machine Learning), que facilita la utilización de estas técnicas por expertos de otros dominios. Este artículo presenta una propuesta concreta de un sistema —AutoGOAL— que ha sido diseñada para resolver problemas de aprendizaje automático de variada naturaleza. Además, se realiza una breve comparación entre sistemas existentes de relevancia en el campo. La propuesta es competitiva con herramientas del estado del arte en problemas clásicos de aprendizaje, a la vez que puede desplegarse, sin esfuerzo adicional, en dominios más complejos, como el procesamiento de lenguaje natural. AutoGOAL constituye un paso más hacia la democratización del aprendizaje automático para usuarios no expertos en el tema., Machine Learning is a field of Artificial Intelligence that has gained recent interest in all areas of the industry, motivated primarily by the accelerated growth of computer capabilities and data availability. However, one of the main difficulties for its application is the need for experts who know the internal details of the multiple models that can be used. In this context, a new field of study has emerged, AutoML (Automated Machine Learning), which facilitates the use of these techniques by experts from other domains. This paper presents a concrete proposal of a system —AutoGOAL— which has been designed to solve machine learning problems of various kinds. In addition, a brief comparison is made between relevant existing systems in the field. The proposal is competitive with state-of-the-art tools in classic machine learning problems, and it can be seamlessly deployed in more complex domains, such as natural language processing. AutoGOAL is another step towards the democratization of machine learning for non-expert users.

El aprendizaje de máquinas ha ganado terreno utilizándose en casi todas las áreas de la vida cotidiana, ayudando a tomar decisiones en las finanzas, la medicina, el comercio y el entretenimiento. El desarrollo continuo de nuevos algoritmos y técnicas de aprendizaje automático, y la amplia gama de herramientas y conjuntos de datos disponibles han traído nuevas oportunidades y desafíos para investigadores y profesionales tanto del mundo académico como de la industria. Seleccionar la mejor estrategia posible para resolver un problema de aprendizaje automático es cada vez más difícil, en parte porque requiere largos tiempos de experimentación y profundos conocimientos técnicos. En este escenario, el campo de investigación Automated Machine Learning (AutoML) ha ganado protagonismo, proponiendo estrategias para automatizar progresivamente tareas usuales durante el desarrollo de aplicaciones de aprendizaje de máquina. Las herramientas de AutoML más comunes permiten seleccionar automáticamente dentro de un conjunto restringido de algoritmos y parámetros la mejor estrategia para cierto conjunto de datos. Sin embargo, los problemas prácticos a menudo requieren combinar y comparar algoritmos heterogéneos implementados con diferentes tecnologías subyacentes. Un ejemplo es el procesamiento del lenguaje natural, un escenario donde varía el espacio de posibles técnicas a aplicar ampliamente entre diferentes tareas, desde el preprocesamiento hasta la representación y clasificación de textos. Realizar AutoML en un escenario heterogéneo como este es complejo porque la solución necesaria podría incluir herramientas y bibliotecas no compatibles entre sí. Esto requeriría que todos los algoritmos acuerden un protocolo común que permita la salida de un algoritmo para ser compartida como entradas a cualquier otro. En esta investigación se diseña e implementa un sistema de AutoML que utiliza técnicas heterogéneas. A diferencia de los enfoques de AutoML existentes, nuestra contribución puede

This paper summarises the eHealth Knowledge Discovery Challenge hosted at IberLEF 2021. We describe the task, resources, and participating systems, highlighting and discussing the main results achieved in the challenge. We analyse the best performing systems and present recommendations for future research., Este artículo resume la Tarea de Descubrimiento de Conocimiento en Salud presentada en IberLEF 2021. Se describen la tarea, los recursos creados, y los sistemas que participaron. Se discuten los resultados principales obtenidos por estos sistemas, y se presentan recomendaciones para continuar la investigación en esta temática.

LETO es un marco de aprendizaje de ontologías diseñado para extraer conocimiento de una variedad de fuentes. Estas fuentes pudieran ser datos estructurados y no estructurados, y de ellas se podrá descubrir, actualizar continuamente, enriquecer e integrar información relevante como parte de un único conocimiento semántico. En la actual versión 1.0 se limita a la extracción de conocimiento desde datos no estructurados, i.e. textos en lenguaje natural, siguiendo el modelo semántico publicado en [EGM2018]. Entre sus funcionalidades está la extracción de entidades y relaciones semánticas desde fuentes textuales; la transformación de esta información en elementos interrelacionados mediante técnicas de agrupamientos; y finalmente generación de ontologías representativas del contenido procesado. Se proporciona un punto de acceso API, y una herramienta visual para la manipulación de procesos y visualización de las ontologías obtenidas [EMA2019]., LETO is an ontology learning framework designed to extract knowledge from a variety of sources. These sources may be structured and/or unstructured data, and from them we can discover, continuously update, enrich and integrate relevant information as part of a single semantic knowledge resource. The current 1.0 version is limited to the extraction of knowledge from unstructured data, i.e. natural language texts, following the semantic model published in [EGM2018]. Among this version’s functionalities are the extraction of entities and semantic relations from textual sources; the transformation of such information into linked elements through clustering techniques; and finally, the generation of representative ontologies of the processed content. An API access point as well as a visual tool for the manipulation of processes and visualization of the obtained ontologies is provided [EMA2019].

This paper introduces Hierarchical Machine Learning Optimisation (HML-Opt), an AutoML framework that is based on probabilistic grammatical evolution. HML-Opt has been designed to provide a flexible framework where a researcher can define the space of possible pipelines to solve a specific machine learning problem, which can range from high-level decisions about representation and features to low-level hyper-parameter values. The evaluation of HML-Opt is presented via two different case studies, both of which demonstrate that it is competitive with existing AutoML tools on a variety of benchmarks. Furthermore, HML-Opt can be applied to novel problems, such as knowledge extraction from natural language text, whereas other techniques are insufficiently flexible to capture the complexity of these scenarios. The source code for HML-Opt is available online for the research community.

This paper presents AutoGOAL, a system for automatic machine learning (AutoML) that uses heterogeneous techniques. In contrast with existing AutoML approaches, our contribution can automatically build machine learning pipelines that combine techniques and algorithms from different frameworks, including shallow classifiers, natural language processing tools, and neural networks. We define the heterogeneous AutoML optimization problem as the search for the best sequence of algorithms that transforms specific input data into the desired output. This provides a novel theoretical and practical approach to AutoML. Our proposal is experimentally evaluated in diverse machine learning problems and compared with alternative approaches, showing that it is competitive with other AutoML alternatives in standard benchmarks. Furthermore, it can be applied to novel scenarios, such as several NLP tasks, where existing alternatives cannot be directly deployed. The system is freely available and includes in-built compatibility with a large number of popular machine learning frameworks, which makes our approach useful for solving practical problems with relative ease and effort.

This paper introduces a web demo that showcases the main characteristics of the AutoGOAL framework. AutoGOAL is a framework in Python for automatically finding the best way to solve a given task. It has been designed mainly for automatic machine learning (AutoML) but it can be used in any scenario where several possible strategies are available to solve a given computational task. In contrast with alternative frameworks, AutoGOAL can be applied seamlessly to Natural Language Processing as well as structured classification problems. This paper presents an overview of the framework’s design and experimental evaluation in several machine learning problems, including two recent NLP challenges. The accompanying software demo is available online and full source code is provided under the MIT open-source license.

This paper presents the preliminary results of an ongoing project that analyzes the growing body of scientific research published around the COVID-19 pandemic. In this research, a general-purpose semantic model is used to double annotate a batch of 500 sentences that were manually selected by the researchers from the CORD-19 corpus. Afterwards, a baseline text-mining pipeline is designed and evaluated via a large batch of 100,959 sentences. We present a qualitative analysis of the most interesting facts automatically extracted and highlight possible future lines of development. The preliminary results show that general-purpose semantic models are a useful tool for discovering fine-grained knowledge in large corpora of scientific documents.

This paper summarises the results of the third edition of the eHealth Knowledge Discovery (KD) challenge, hosted at the Iberian Language Evaluation Forum 2020. The eHealth-KD challenge proposes two computational tasks involving the identification of semantic entities and relations in natural language text, focusing on Spanish language health documents. In this edition, besides text extracted from medical sources, Wikipedia content was introduced into the corpus, and a novel transfer-learning evaluation scenario was designed that challenges participants to create systems that provide cross-domain generalisation. A total of eight teams participated with a variety of approaches including deep learning end-to-end systems as well as rule-based and knowledge-driven techniques. This paper analyses the most successful approaches and highlights the most interesting challenges for future research in this field.

This paper presents an analysis of Task 3 eHealth-KD challenge in the TASS 2018Workshop. The challenge consisted of the extraction of concepts, actions, and their corresponding semantic relations from health-related documents written in the Spanish language. The documents were manually annotated with a schema based on triples (Subject, Action, Target) and an additional set of semantic relations. Several research teams presented computational systems, obtaining relevant results in different subtasks. In this paper, the approaches performed by each team are analyzed and the most promising lines for future development are highlighted and discussed. Moreover, an in-depth analysis of the results is presented focusing on the main characteristics of each subtask. The overall eHealth-KD analysis has indicated that the Knowledge Discovery (KD) task, specifically focused on concrete domains and languages, represents a rich area for further research. In addition, this study considers that the fusion of machine learning {especially deep learning techniques{ and knowledge-based approaches will benefit the KD task., Este artículo presenta un análisis de la Tarea 3 eHealth-KD in el Taller TASS 2018. La tarea consistió en la extracción de conceptos, acciones, y sus correspondientes relaciones semánticas a partir de documentos sobre temas de salud en idioma español. Los documentos fueron manualmente anotados con un esquema basado en tripletas (Sujeto, Acción, Objeto) y un conjunto adicional de relaciones semánticas. Varios investigadores presentaron sistemas computacionales para la tarea, obteniendo resultados relevantes en las diferentes subtareas definidas. Los enfoques presentados por cada equipo son analizados en este artículo, subrayando las líneas de investigación futura más prometedoras. Además, se presenta un análisis profundo de los resultados, enfocado en las características de cada subtarea. El análisis general de la tarea eHealth-KD indica que las tareas de descubrimiento de conocimiento en idioma español para dominios específicos es un área fructífera de investigación. El progreso en este campo podría beneficiarse considerablemente de la fusión de técnicas de aprendizaje automático -especialmente aprendizaje profundo- con enfoques basados en conocimiento.

This is an overview of the Workshop on Semantic Analysis at the SEPLN congress held in Sevilla, Spain, in September 2018. This forum proposes to participants four different semantic tasks on texts written in Spanish. Task 1 focuses on polarity classification; Task 2 encourages the development of aspect-based polarity classification systems; Task 3 provides a scenario for discovering knowledge from eHealth documents; finally, Task 4 is about automatic classification of news articles according to safety. The former two tasks are novel in this TASS's edition. We detail the approaches and the results of the submitted systems of the different groups in each task., Este artículo ofrece un resumen sobre el Taller de Análisis Semántico en la SEPLN (TASS) celebrado en Sevilla, España, en septiembre de 2018. Este foro propone a los participantes cuatro tareas diferentes de análisis semántico sobre textos en español. La Tarea 1 se centra en la clasificación de la polaridad; la Tarea 2 anima al desarrollo de sistemas de polaridad orientados a aspectos; la Tarea 3 consiste en descubrir conocimiento en documentos sobre salud; finalmente, la Tarea 4 propone la clasificación automática de noticias periodísticas según un nivel de seguridad. Las dos últimas tareas son nuevas en esta edición. Se ofrece una síntesis de los sistemas y los resultados aportados por los distintos equipos participantes, así como una discusión sobre los mismos.

During the search process of differential evolution (DE), each new solution may represent a new more promising region of the search space (exploration) or a better solution within the current region (exploitation). This concurrent exploitation can interfere with exploration since the identification of a new more promising region depends on finding a (random) solution in that region which is better than its target solution. Ideally, every sampled solution will have the same relative fitness with respect to its nearby local optimum – finding the best region to exploit then becomes the problem of finding the best random solution. However, differential evolution is characterized by an initial period of exploration followed by rapid convergence. Once the population starts converging, the difference vectors become shorter, more exploitation is performed, and an accelerating convergence occurs. This rapid convergence can occur well before the algorithm’s budget of function evaluations is exhausted; that is, the algorithm can converge prematurely. In thresheld convergence, early exploitation is “held” back by a threshold function, allowing a longer exploration phase. This paper presents a new adaptive thresheld convergence mechanism which helps DE achieve large performance improvements in multi-modal search spaces.