http://sedici.unlp.edu.ar:80/handle/10915/75936 Tue, 10 Nov 2026 02:36:56 GMT 2026-11-10T02:36:56Z http://sedici.unlp.edu.ar:80/handle/10915/76361 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). El presente trabajo de investigación se propone reformular las condiciones para una definición de una función de contracción Horn que esta basada en la noción de importancia epistómica (EE: epistemic entrenchment) que es una construcción de contracción equivalente al modelo AGM. Dicha construcción considera disyunciones arbitrarias, las cuales no son aptas para entornos con lógicas Horn. Por ello es que se hace uso de una técnica de aproximación Horn conocida como Horn Strengthening. Es por ello que en este artículo se reformulan las condiciones para definir la función de contracción Horn basada en EE, y se prueba la validez de estas nuevas condiciones evaluóndolas con el conjunto de postulados que caracterizan este tipo de operacion, demostrando que la funcion de contracción Horn satisface ciertos postulados característicos. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76361 2013-01-01T00:00:00Z El presente trabajo de investigación se propone reformular las condiciones para una definición de una función de contracción Horn que esta basada en la noción de importancia epistómica (EE: epistemic entrenchment) que es una construcción de contracción equivalente al modelo AGM. Dicha construcción considera disyunciones arbitrarias, las cuales no son aptas para entornos con lógicas Horn. Por ello es que se hace uso de una técnica de aproximación Horn conocida como Horn Strengthening. Es por ello que en este artículo se reformulan las condiciones para definir la función de contracción Horn basada en EE, y se prueba la validez de estas nuevas condiciones evaluóndolas con el conjunto de postulados que caracterizan este tipo de operacion, demostrando que la funcion de contracción Horn satisface ciertos postulados característicos. http://sedici.unlp.edu.ar:80/handle/10915/76359 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). We present a method for enlarge a lexicon (with frequencies information), that is useful for parsing and others NLP applications. We show an example enlarging the verbal LGLex lexicon of French [8], using several corpora extracted from the evaluation campaign for French parsers Passage [5]. To do that, we use the results of the frmg parser [7] with IRASubcat, a tool that automatically acquires subcategorization frames from corpus in any language and that also allows to complete an existing lexicon. We obtain the frequencies of occurrence for each input and each subcategorization frame for 14,068 distinct lemmas. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76359 2013-01-01T00:00:00Z We present a method for enlarge a lexicon (with frequencies information), that is useful for parsing and others NLP applications. We show an example enlarging the verbal LGLex lexicon of French [8], using several corpora extracted from the evaluation campaign for French parsers Passage [5]. To do that, we use the results of the frmg parser [7] with IRASubcat, a tool that automatically acquires subcategorization frames from corpus in any language and that also allows to complete an existing lexicon. We obtain the frequencies of occurrence for each input and each subcategorization frame for 14,068 distinct lemmas. http://sedici.unlp.edu.ar:80/handle/10915/76358 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). Latent Semantic Analysis is a natural language processing tools that allows estimating semantic distance between terms. The success of LSA is mainly based on the training corpus choice, which have been studied principally in English. This study focuses on studying LSA with regional Spanish corpus and evaluate the performance by identifying synonyms. We found that performance was slightly better than chance, concordantly with previous results. Standard LSA method cannot dynamically increase the training corpus. By using classifiers we combined multiple LSA models and showed that the use of automatic classifiers increase the performance. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76358 2013-01-01T00:00:00Z Latent Semantic Analysis is a natural language processing tools that allows estimating semantic distance between terms. The success of LSA is mainly based on the training corpus choice, which have been studied principally in English. This study focuses on studying LSA with regional Spanish corpus and evaluate the performance by identifying synonyms. We found that performance was slightly better than chance, concordantly with previous results. Standard LSA method cannot dynamically increase the training corpus. By using classifiers we combined multiple LSA models and showed that the use of automatic classifiers increase the performance. http://sedici.unlp.edu.ar:80/handle/10915/76356 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). We present SuFLexQA, a system for Question Answering that integrates deep linguistic information from verbal lexica into Quepy, a generic framework for translating natural language questions into a query language. We are participating in the QALD-3 contest to assess the main achievements and shortcomings of the system. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76356 2013-01-01T00:00:00Z We present SuFLexQA, a system for Question Answering that integrates deep linguistic information from verbal lexica into Quepy, a generic framework for translating natural language questions into a query language. We are participating in the QALD-3 contest to assess the main achievements and shortcomings of the system. http://sedici.unlp.edu.ar:80/handle/10915/76355 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). La conformación de comisiones es una actividad en la que se requiere seleccionar un conjunto de candidatos a partir de una población de individuos. Estos individuos y sus relaciones pueden representarse mediante una red social. Trabajos previos relacionados con el análisis de redes sociales se enfocan en la identificación de la importancia de los nodos que conforman esas redes, o bien en las propiedades estructurales de las mismas. Este trabajo propone una alternativa para la selección de un conjunto de individuos que presenten menor relación entre sí a partir de una red social, considerando esto como un criterio de independencia entre los mismos. El enfoque propuesto es evaluado con un caso de estudio en el cual se propone en primer lugar, la construcción de una red social en base a datos de coautoría de publicaciones de investigadores y mismo lugar de trabajo, la cual es analizada para determinar las distancias entre cada par de individuos de la red, y en segundo lugar, la optimización de las distancias entre los posibles candidatos mediante un algoritmo genético. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76355 2013-01-01T00:00:00Z La conformación de comisiones es una actividad en la que se requiere seleccionar un conjunto de candidatos a partir de una población de individuos. Estos individuos y sus relaciones pueden representarse mediante una red social. Trabajos previos relacionados con el análisis de redes sociales se enfocan en la identificación de la importancia de los nodos que conforman esas redes, o bien en las propiedades estructurales de las mismas. Este trabajo propone una alternativa para la selección de un conjunto de individuos que presenten menor relación entre sí a partir de una red social, considerando esto como un criterio de independencia entre los mismos. El enfoque propuesto es evaluado con un caso de estudio en el cual se propone en primer lugar, la construcción de una red social en base a datos de coautoría de publicaciones de investigadores y mismo lugar de trabajo, la cual es analizada para determinar las distancias entre cada par de individuos de la red, y en segundo lugar, la optimización de las distancias entre los posibles candidatos mediante un algoritmo genético. http://sedici.unlp.edu.ar:80/handle/10915/76353 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). Making Decision in non-structured problems is not a simple task. For this reason, decision makers use Decision Support Systems (DSS). These kinds of systems implement techniques and algorithms in order to improve the decision process. Analytic Hierarchy Process (AHP) is one of these techniques which yields a ranking scale of alternatives based on criteria and alternative matrices. These matrices make a pairwise comparison between a set of elements compared and must be complete and consistent in order to be processed with AHP. Incompleteness and inconsistency emerge as a consequence of the large data required to be compared by an expert, which exceeds his or her human abilities. Genetic Algorithms (GA) is a powerful used technique which provides simplicity, broad applicability and flexibility for search problems. In this work a GA model is exposed, being its aims to help the expert to fill the matrix and provide reasonable judgments by suggesting possible values. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76353 2013-01-01T00:00:00Z Making Decision in non-structured problems is not a simple task. For this reason, decision makers use Decision Support Systems (DSS). These kinds of systems implement techniques and algorithms in order to improve the decision process. Analytic Hierarchy Process (AHP) is one of these techniques which yields a ranking scale of alternatives based on criteria and alternative matrices. These matrices make a pairwise comparison between a set of elements compared and must be complete and consistent in order to be processed with AHP. Incompleteness and inconsistency emerge as a consequence of the large data required to be compared by an expert, which exceeds his or her human abilities. Genetic Algorithms (GA) is a powerful used technique which provides simplicity, broad applicability and flexibility for search problems. In this work a GA model is exposed, being its aims to help the expert to fill the matrix and provide reasonable judgments by suggesting possible values. http://sedici.unlp.edu.ar:80/handle/10915/76352 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). Argumentation is an attractive reasoning mechanism due to its dialectical and non monotonic nature, and its properties of computational tractability. In dynamic domains where the agents deal with incomplete and contradictory information, to determine the accepted or warranted information, an argument comparison criterion must be used. Argumentation systems that use a single argument comparison criterion have been widely studied in the literature. In some of these approaches, the comparison is xed and in others the criterion can be replaced in a modular way. In this work we introduce an argumentative server that provides recommendations to its client agents and the ability to decide how multiple argument comparison criteria can be combined. In the proposed formalism, the argumentative reasoning is based on the criteria selected by the client agents. As a result, a set of operators to combine multiple preference criteria is presented. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76352 2013-01-01T00:00:00Z Argumentation is an attractive reasoning mechanism due to its dialectical and non monotonic nature, and its properties of computational tractability. In dynamic domains where the agents deal with incomplete and contradictory information, to determine the accepted or warranted information, an argument comparison criterion must be used. Argumentation systems that use a single argument comparison criterion have been widely studied in the literature. In some of these approaches, the comparison is xed and in others the criterion can be replaced in a modular way. In this work we introduce an argumentative server that provides recommendations to its client agents and the ability to decide how multiple argument comparison criteria can be combined. In the proposed formalism, the argumentative reasoning is based on the criteria selected by the client agents. As a result, a set of operators to combine multiple preference criteria is presented. http://sedici.unlp.edu.ar:80/handle/10915/76349 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). Los sistemas de Aprendizaje Colaborativo Soportado por Computadora (ACSC) permiten el aprendizaje grupal con independencia del tiempo y espacio donde estén localizados los estudiantes y los docentes. Sin embargo, las interacciones que conducen a la creación colaborativa de conocimiento no surgen de manera espontánea y la tecnología puede inhibirlas u obstaculizarlas. Para colaborar efectivamente los estudiantes necesitan principalmente de un etutor (docente) que coordine la interacción grupal. La selección de e-tutores capacitados es clave para el éxito del ACSC pero el análisis manual de las interacciones registradas en estos entornos para conocer el desempeño de los docentes requiere mucho tiempo y esfuerzo. En este artículo se describe un trabajo de investigación que aplica técnicas de minería de textos para crear clasificadores que permitan identificar automáticamente las habilidades manifestadas por e-tutores. Los resultados obtenidos mediante diferentes algoritmos de clasificación son presentados, analizados y comparados. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76349 2013-01-01T00:00:00Z Los sistemas de Aprendizaje Colaborativo Soportado por Computadora (ACSC) permiten el aprendizaje grupal con independencia del tiempo y espacio donde estén localizados los estudiantes y los docentes. Sin embargo, las interacciones que conducen a la creación colaborativa de conocimiento no surgen de manera espontánea y la tecnología puede inhibirlas u obstaculizarlas. Para colaborar efectivamente los estudiantes necesitan principalmente de un etutor (docente) que coordine la interacción grupal. La selección de e-tutores capacitados es clave para el éxito del ACSC pero el análisis manual de las interacciones registradas en estos entornos para conocer el desempeño de los docentes requiere mucho tiempo y esfuerzo. En este artículo se describe un trabajo de investigación que aplica técnicas de minería de textos para crear clasificadores que permitan identificar automáticamente las habilidades manifestadas por e-tutores. Los resultados obtenidos mediante diferentes algoritmos de clasificación son presentados, analizados y comparados. http://sedici.unlp.edu.ar:80/handle/10915/76222 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). This paper studies alternatives to solve the problem of autonomous mobile robots navigation in unknown indoor environments. The navigation system uses fuzzy logic to combine the information obtained from range sensors and the navigational data to plan the robot’s movements. The strategy is built upon five modules: i) target following, ii) obstacle avoidance, iii) possible path, iv) deadlock detection and v) wall following. Given a possible path and obstacles near the environment of the robot, the controller will modulate the output velocity in order to go to the target and avoid collisions. In case of dead lock situations, a method that enables the robot to detect, escape and reach the target is proposed. The performance and behavior of the proposed navigational system was evaluated through simulations in different conditions, where the effectiveness of the proposed method is demonstrated and compared with previous results. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76222 2013-01-01T00:00:00Z This paper studies alternatives to solve the problem of autonomous mobile robots navigation in unknown indoor environments. The navigation system uses fuzzy logic to combine the information obtained from range sensors and the navigational data to plan the robot’s movements. The strategy is built upon five modules: i) target following, ii) obstacle avoidance, iii) possible path, iv) deadlock detection and v) wall following. Given a possible path and obstacles near the environment of the robot, the controller will modulate the output velocity in order to go to the target and avoid collisions. In case of dead lock situations, a method that enables the robot to detect, escape and reach the target is proposed. The performance and behavior of the proposed navigational system was evaluated through simulations in different conditions, where the effectiveness of the proposed method is demonstrated and compared with previous results. http://sedici.unlp.edu.ar:80/handle/10915/76220 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). Clustering is fundamental to understand the structure of data. In the past decade the cluster ensemble problem has been introduced, which combines a set of partitions (an ensemble) of the data to obtain a single consensus solution that outperforms all the ensemble members. Although disagreement among ensemble partitions (diversity) has been found to be fundamental for success, the literature has arrived to confusing conclusions: some authors suggest that high diversity is beneficial for the final performance, whereas others have indicated that medium is better. While there are several options to measure the diversity, there is no method to control it. This paper introduces a new ensemble generation strategy and a method to smoothly change the ensemble diversity. Experimental results on three datasets suggest that this is an important step towards a more systematic approach to analyze the impact of the ensemble diversity on the overall consensus performance. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76220 2013-01-01T00:00:00Z Clustering is fundamental to understand the structure of data. In the past decade the cluster ensemble problem has been introduced, which combines a set of partitions (an ensemble) of the data to obtain a single consensus solution that outperforms all the ensemble members. Although disagreement among ensemble partitions (diversity) has been found to be fundamental for success, the literature has arrived to confusing conclusions: some authors suggest that high diversity is beneficial for the final performance, whereas others have indicated that medium is better. While there are several options to measure the diversity, there is no method to control it. This paper introduces a new ensemble generation strategy and a method to smoothly change the ensemble diversity. Experimental results on three datasets suggest that this is an important step towards a more systematic approach to analyze the impact of the ensemble diversity on the overall consensus performance. http://sedici.unlp.edu.ar:80/handle/10915/76216 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). La minería de textos tiene un importante potencial, ya que gran parte de la información de las organizaciones está disponible en documentos de texto u otra información no estructurada. Una de las tareas integrales de la minería de textos es la extracción de entidades con nombre (NER). El presente trabajo describe los principales enfoques en uso para esta tarea y los aplica a un problema concreto, la extracción de información de un corpus de 8000 documentos correspondientes a resoluciones rectorales. Los experimentos muestran que los campos aleatorios condicionales (CRFs) son la técnica más adecuada para este problema. El trabajo describe también la arquitectura para la gestión de información no estructurada en la que se enmarca esta tarea y de la que forma parte. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76216 2013-01-01T00:00:00Z La minería de textos tiene un importante potencial, ya que gran parte de la información de las organizaciones está disponible en documentos de texto u otra información no estructurada. Una de las tareas integrales de la minería de textos es la extracción de entidades con nombre (NER). El presente trabajo describe los principales enfoques en uso para esta tarea y los aplica a un problema concreto, la extracción de información de un corpus de 8000 documentos correspondientes a resoluciones rectorales. Los experimentos muestran que los campos aleatorios condicionales (CRFs) son la técnica más adecuada para este problema. El trabajo describe también la arquitectura para la gestión de información no estructurada en la que se enmarca esta tarea y de la que forma parte. http://sedici.unlp.edu.ar:80/handle/10915/76214 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). Many real-world problems may vary over time. These non stationary problems have been widely studied in the literature, often called drifting concepts problems. Recently, deep architectures have drawn a growing attention, given that they can easily model functions that are hard to approximate with shallow ones and an effective way of training them have been discovered. In this work we adapt a deep architecture to problems that present concept drift. To this end we show a way of combining them with a widely known drifting concept technique, the Streaming Ensemble Algorithm. We evaluate the new method using appropriate drifting problems and compare its performance with a more traditional approach. The results obtained are promising and show that the proposed variation is effective at combining the expressive power of a deep architecture with the adaptability of SEA. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76214 2013-01-01T00:00:00Z Many real-world problems may vary over time. These non stationary problems have been widely studied in the literature, often called drifting concepts problems. Recently, deep architectures have drawn a growing attention, given that they can easily model functions that are hard to approximate with shallow ones and an effective way of training them have been discovered. In this work we adapt a deep architecture to problems that present concept drift. To this end we show a way of combining them with a widely known drifting concept technique, the Streaming Ensemble Algorithm. We evaluate the new method using appropriate drifting problems and compare its performance with a more traditional approach. The results obtained are promising and show that the proposed variation is effective at combining the expressive power of a deep architecture with the adaptability of SEA. http://sedici.unlp.edu.ar:80/handle/10915/76213 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). When applying clustering algorithms on biological data the information about biological processes is not usually present in an explicit way, although this knowledge is later used by biologists to validate the clusters and the relations found among data. This work presents a new distance measure for biological data which combines expression and semantic information, in order to be used into a clustering algorithm. The distance is calculated pairwise among all pairs of genes and it is incorporated during the training process of the clustering algorithm. The approach was evaluated on two real datasets using several validation measures. The obtained results are consistent across all the measures, showing better semantic quality for clusters with the new algorithm in comparison to standard clustering. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76213 2013-01-01T00:00:00Z When applying clustering algorithms on biological data the information about biological processes is not usually present in an explicit way, although this knowledge is later used by biologists to validate the clusters and the relations found among data. This work presents a new distance measure for biological data which combines expression and semantic information, in order to be used into a clustering algorithm. The distance is calculated pairwise among all pairs of genes and it is incorporated during the training process of the clustering algorithm. The approach was evaluated on two real datasets using several validation measures. The obtained results are consistent across all the measures, showing better semantic quality for clusters with the new algorithm in comparison to standard clustering. http://sedici.unlp.edu.ar:80/handle/10915/76211 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). The nature has inspired several metaheuristics, outstanding among these is Ant Colony Optimization (ACO), which have proved to be very effective and efficient in problems of high complexity (NP-hard) in combinatorial optimization. This paper describes the implementation of an ACO model algorithm known as Elitist Ant System (EAS), applied to a combinatorial optimization problem called Job Shop Scheduling Problem (JSSP). We propose a method that seeks to reduce delays designating the operation immediately available, but considering the operations that lack little to be available and have a greater amount of pheromone. The performance of the algorithm was evaluated for problems of JSSP reference, comparing the quality of the solutions obtained regarding the best known solution of the most effective methods. The solutions were of good quality and obtained with a remarkable efficiency by having to make a very low number of objective function evaluations. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76211 2013-01-01T00:00:00Z The nature has inspired several metaheuristics, outstanding among these is Ant Colony Optimization (ACO), which have proved to be very effective and efficient in problems of high complexity (NP-hard) in combinatorial optimization. This paper describes the implementation of an ACO model algorithm known as Elitist Ant System (EAS), applied to a combinatorial optimization problem called Job Shop Scheduling Problem (JSSP). We propose a method that seeks to reduce delays designating the operation immediately available, but considering the operations that lack little to be available and have a greater amount of pheromone. The performance of the algorithm was evaluated for problems of JSSP reference, comparing the quality of the solutions obtained regarding the best known solution of the most effective methods. The solutions were of good quality and obtained with a remarkable efficiency by having to make a very low number of objective function evaluations. http://sedici.unlp.edu.ar:80/handle/10915/76153 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). This paper addresses the image modeling problem under the assumption that images can be represented by 2d order, hidden Markov random fields models. The modeling applications we have in mind com- prise pixelwise segmentation of gray-level images coming from the field of Oral Radiographic Differential Diagnosis. Segmentation is achieved by approximations to the solution of the maximum a posteriori equation (MAP) when the emission distribution is assumed the same in all models and the difference lays in the Neighborhood Markovian hypothesis made over the labeling random field. For two algorithms, 2d path-constrained Viterbi training and Potts-ICM we investigate goodness of fit by study- ing statistical complexity, computational gain, extent of automation, and rate of classification measured with kappa statistic. All code written is provided in a Matlab toolbox available for download from our website, following the Reproducible Research Paradigm. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76153 2013-01-01T00:00:00Z This paper addresses the image modeling problem under the assumption that images can be represented by 2d order, hidden Markov random fields models. The modeling applications we have in mind com- prise pixelwise segmentation of gray-level images coming from the field of Oral Radiographic Differential Diagnosis. Segmentation is achieved by approximations to the solution of the maximum a posteriori equation (MAP) when the emission distribution is assumed the same in all models and the difference lays in the Neighborhood Markovian hypothesis made over the labeling random field. For two algorithms, 2d path-constrained Viterbi training and Potts-ICM we investigate goodness of fit by study- ing statistical complexity, computational gain, extent of automation, and rate of classification measured with kappa statistic. All code written is provided in a Matlab toolbox available for download from our website, following the Reproducible Research Paradigm. http://sedici.unlp.edu.ar:80/handle/10915/76150 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). Everyday human beings are faced with situations they should choose among different alternatives by means of reasoning and mental processes when solving a problem. Many of these decision problems are under uncertain environments including vague, imprecise and subjective information that is usually modeled by linguistic information due to the use of natural language and its relation to mental reasoning processes of the experts when expressing their judgments. In a decision process multiple criteria can be evaluated which involving multiple experts with different degrees of knowledge. Such process can be modeled by using Multi-granular Linguistic Information (MGLI) and Computing with Words (CW) processes to solve the related decision problems. Different methodologies and approaches have been proposed to accomplish this process in an accurate and interpretable way. In this paper we propose a useful Decision Analysis Framework to manage this kind of problems by using the Extended Linguistic Hierarchy (ELH), 2-tuples linguistic representation model and its computational method. The developed Framework has many advantages when dealing with a complex problem in a simple way and its capability of having easy and useful reasonably results. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76150 2013-01-01T00:00:00Z Everyday human beings are faced with situations they should choose among different alternatives by means of reasoning and mental processes when solving a problem. Many of these decision problems are under uncertain environments including vague, imprecise and subjective information that is usually modeled by linguistic information due to the use of natural language and its relation to mental reasoning processes of the experts when expressing their judgments. In a decision process multiple criteria can be evaluated which involving multiple experts with different degrees of knowledge. Such process can be modeled by using Multi-granular Linguistic Information (MGLI) and Computing with Words (CW) processes to solve the related decision problems. Different methodologies and approaches have been proposed to accomplish this process in an accurate and interpretable way. In this paper we propose a useful Decision Analysis Framework to manage this kind of problems by using the Extended Linguistic Hierarchy (ELH), 2-tuples linguistic representation model and its computational method. The developed Framework has many advantages when dealing with a complex problem in a simple way and its capability of having easy and useful reasonably results. http://sedici.unlp.edu.ar:80/handle/10915/76148 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). En este artículo se presenta la investigación, evaluación y comparación experimental de técnicas de procesamiento de lenguaje natural para análisis de información subjetiva como opiniones, sentimientos y emociones en textos no estructurados en idioma español. Se implemento una herramienta para la extracción de opiniones de documentos con el objetivo de clasificarlos según polaridad de sentimientos (positivos o negativos) utilizando como corpus de datos la base de comentarios de un sitio de crítica gastronómica al que se le aplican distintas técnicas de preprocesamiento. La principal contribución de este artículo es la experimentación y evaluación de tecnificas de clasificación subjetiva de textos para el idioma español en función del tamaño de corpus, tipos de atributos extraídos y preprocesamientos aplicados. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76148 2013-01-01T00:00:00Z En este artículo se presenta la investigación, evaluación y comparación experimental de técnicas de procesamiento de lenguaje natural para análisis de información subjetiva como opiniones, sentimientos y emociones en textos no estructurados en idioma español. Se implemento una herramienta para la extracción de opiniones de documentos con el objetivo de clasificarlos según polaridad de sentimientos (positivos o negativos) utilizando como corpus de datos la base de comentarios de un sitio de crítica gastronómica al que se le aplican distintas técnicas de preprocesamiento. La principal contribución de este artículo es la experimentación y evaluación de tecnificas de clasificación subjetiva de textos para el idioma español en función del tamaño de corpus, tipos de atributos extraídos y preprocesamientos aplicados. http://sedici.unlp.edu.ar:80/handle/10915/76139 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). Analizar la estabilidad de una solución de clustering implica medir la capacidad de un algoritmo para producir resultados similares dada una misma fuente de datos de entrada. Los índices de validación externa permiten cuantificar dicha similitud entre un par de soluciones de clustering. Dentro de los índices clásicos mas utilizados es posible validar soluciones con clusters no solapados, en donde cada patrón sólo puede pertenecer a un cluster. Sin embargo, en aplicaciones prácticas, generalmente se dan situaciones en las que un patrón podría poseer más de una etiqueta. En este trabajo se analiza un índice de validación externa desde un enfoque probabilístico y se provee una reformulación aplicable a soluciones con clusters solapados. Luego de presentar el nuevo índice, se muestran y discuten resultados de experimentos realizados sobre ejemplos artificiales y una base de datos real. Los resultados muestran cómo el nuevo índice puede medir adecuadamente la similitud entre clusters solapados, permitiendo asó analizar la estabilidad en ambos casos. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/76139 2013-01-01T00:00:00Z Analizar la estabilidad de una solución de clustering implica medir la capacidad de un algoritmo para producir resultados similares dada una misma fuente de datos de entrada. Los índices de validación externa permiten cuantificar dicha similitud entre un par de soluciones de clustering. Dentro de los índices clásicos mas utilizados es posible validar soluciones con clusters no solapados, en donde cada patrón sólo puede pertenecer a un cluster. Sin embargo, en aplicaciones prácticas, generalmente se dan situaciones en las que un patrón podría poseer más de una etiqueta. En este trabajo se analiza un índice de validación externa desde un enfoque probabilístico y se provee una reformulación aplicable a soluciones con clusters solapados. Luego de presentar el nuevo índice, se muestran y discuten resultados de experimentos realizados sobre ejemplos artificiales y una base de datos real. Los resultados muestran cómo el nuevo índice puede medir adecuadamente la similitud entre clusters solapados, permitiendo asó analizar la estabilidad en ambos casos. http://sedici.unlp.edu.ar:80/handle/10915/75956 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). Group formation is a key aspect in computer-supported collaborative learning, since different characteristics of students might influence the group performance. In this article, we present an assistant that models group formation as a weighted constraint satisfaction problem (WCSP), and considers three students’ features, namely: psychological styles, team roles and social networks. Our WCSP formulation is able to combine constraints and preferences for individuals and groups. This assistant can aid teachers to form groups considering factors such as team role balance and distribution of psychological styles. We report on a pilot study to evaluate the proposal in different scenarios. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/75956 2013-01-01T00:00:00Z Group formation is a key aspect in computer-supported collaborative learning, since different characteristics of students might influence the group performance. In this article, we present an assistant that models group formation as a weighted constraint satisfaction problem (WCSP), and considers three students’ features, namely: psychological styles, team roles and social networks. Our WCSP formulation is able to combine constraints and preferences for individuals and groups. This assistant can aid teachers to form groups considering factors such as team role balance and distribution of psychological styles. We report on a pilot study to evaluate the proposal in different scenarios. http://sedici.unlp.edu.ar:80/handle/10915/75954 Objeto de conferencia XIV Argentine Symposium on Artificial Intelligence (ASAI) - JAIIO 42 (2013). El uso de plataformas colaborativas en los procesos de enseñanza es muy usual hoy en día. Las mismas facilitan la comunicación entre docentes y estudiantes que pueden estar distantes físicamente. Pero uno de los problemas detectados son los nuevos usuarios, no saben quién es la persona más idónea para comenzar a interactuar sobre alguna temática, muchas veces las interacciones no son exitosas debido a que la persona con quien interactuó resulta no ser la idónea. En este artículo se presenta un sistema recomendador que recomienda a una persona un usuario idóneo con quien interactuar. Para demostrar la validez de este sistema recomendador se realizaron experimentos cuyos resultados demuestran que usuarios que interactuaron con usuarios recomendados de forma personalizada (en base a su perfil y la reputación del mismo dentro la plataforma colaborativa) fueron más exitosas. Tue, 01 Jan 2013 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/75954 2013-01-01T00:00:00Z El uso de plataformas colaborativas en los procesos de enseñanza es muy usual hoy en día. Las mismas facilitan la comunicación entre docentes y estudiantes que pueden estar distantes físicamente. Pero uno de los problemas detectados son los nuevos usuarios, no saben quién es la persona más idónea para comenzar a interactuar sobre alguna temática, muchas veces las interacciones no son exitosas debido a que la persona con quien interactuó resulta no ser la idónea. En este artículo se presenta un sistema recomendador que recomienda a una persona un usuario idóneo con quien interactuar. Para demostrar la validez de este sistema recomendador se realizaron experimentos cuyos resultados demuestran que usuarios que interactuaron con usuarios recomendados de forma personalizada (en base a su perfil y la reputación del mismo dentro la plataforma colaborativa) fueron más exitosas.