http://sedici.unlp.edu.ar:80/handle/10915/62791 Mon, 02 Nov 2026 22:16:03 GMT 2026-11-02T22:16:03Z http://sedici.unlp.edu.ar:80/handle/10915/63287 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) In these days, there are a growing interest in pattern recognition for tasks as predicting weather events, recommending best routes, intrusion detection or face detection. These tasks can be modelled as a classification problem, where a common alternative is using an ensemble model of classification. An usual ensemble model is given by Mixture of Experts model, which belongs to modular artificial neural networks consisting of two subcomponents type: networks of experts and Gating network, and whose combination creates an environment of competition among experts seeking to obtain patterns of the data source, in order to specialize in that particular task, all this supervised in the Gating network, which is the mediator agent and ponders the quality delivered by each expert model solution. We observe that this architecture assume that one gate influence one data point, consequently the training can be misleading to real datasets where the data is better explained by multiple experts. In this work, we present a variant of traditional MoE model, which consists of maximizing the entropy of the evaluation function in the Gating network in conjunction with standard error minimization. The results show the advantage of our approach in multiple datasets in terms of accuracy metric. As a future work, we plan to apply this idea to the Mixture-of-Experts with embedded feature selection. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63287 2017-01-01T00:00:00Z In these days, there are a growing interest in pattern recognition for tasks as predicting weather events, recommending best routes, intrusion detection or face detection. These tasks can be modelled as a classification problem, where a common alternative is using an ensemble model of classification. An usual ensemble model is given by Mixture of Experts model, which belongs to modular artificial neural networks consisting of two subcomponents type: networks of experts and Gating network, and whose combination creates an environment of competition among experts seeking to obtain patterns of the data source, in order to specialize in that particular task, all this supervised in the Gating network, which is the mediator agent and ponders the quality delivered by each expert model solution. We observe that this architecture assume that one gate influence one data point, consequently the training can be misleading to real datasets where the data is better explained by multiple experts. In this work, we present a variant of traditional MoE model, which consists of maximizing the entropy of the evaluation function in the Gating network in conjunction with standard error minimization. The results show the advantage of our approach in multiple datasets in terms of accuracy metric. As a future work, we plan to apply this idea to the Mixture-of-Experts with embedded feature selection. http://sedici.unlp.edu.ar:80/handle/10915/63286 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) The advance of technology makes that mobile devices have gained widespread popularity. Modern mobile devices include build in sensors, cameras, compasses, and enhanced storage and processing capabilities, which allow developers to use those features to create applications with new or enhanced functionality. In this context we present a mobile application for searching places, people an events within a university campus. In our work we leverage semantic web and augmented reality to provide an application with a high degree of query expressiveness and an enhanced user experience. In addition, we validate our approach with a use case example that shows the complete searching process. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63286 2017-01-01T00:00:00Z The advance of technology makes that mobile devices have gained widespread popularity. Modern mobile devices include build in sensors, cameras, compasses, and enhanced storage and processing capabilities, which allow developers to use those features to create applications with new or enhanced functionality. In this context we present a mobile application for searching places, people an events within a university campus. In our work we leverage semantic web and augmented reality to provide an application with a high degree of query expressiveness and an enhanced user experience. In addition, we validate our approach with a use case example that shows the complete searching process. http://sedici.unlp.edu.ar:80/handle/10915/63267 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) Data ambiguity from various sources remains as a complex problem that affects services provided by digital libraries. From the point of view of integration of information from different sources, the challenge of author ambiguity is one of the most important, and there are numerous methods proposed to deal with this issue using different approaches. They generally work for some scenarios but they have important limitations, specially when dealing with heterogeneous sources. In this work, we review a group of existing methods and then propose a technique that combines some of them, also incorporating a measure of distance using semantic technologies to solve the ambiguity of authors while integrating bibliographic data from various sources. This technique has been successfully tested in disambiguating Ecuadorian authors from both internal sources (institutional repositories) and external digital libraries. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63267 2017-01-01T00:00:00Z Data ambiguity from various sources remains as a complex problem that affects services provided by digital libraries. From the point of view of integration of information from different sources, the challenge of author ambiguity is one of the most important, and there are numerous methods proposed to deal with this issue using different approaches. They generally work for some scenarios but they have important limitations, specially when dealing with heterogeneous sources. In this work, we review a group of existing methods and then propose a technique that combines some of them, also incorporating a measure of distance using semantic technologies to solve the ambiguity of authors while integrating bibliographic data from various sources. This technique has been successfully tested in disambiguating Ecuadorian authors from both internal sources (institutional repositories) and external digital libraries. http://sedici.unlp.edu.ar:80/handle/10915/63265 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) A importância de trazer dados do modelo relacional para outros modelos e tecnologias tem sido amplamente debatidos, como por exemplo a publicação de dados como grafos. Este modelo permite executar análises topológicas, tal como ocorre nas análises de redes sociais, predição de ligações e sistemas de recomendação. Existem iniciativas para mapear de um banco de dados relacional para a representação em grafo. No entanto, eles não consideram as diferentes maneiras de gerar esses grafos, especialmente quando o objetivo é realizar análises topológicas. Este trabalho propõe heurísticas para facilitar a sistematização do mapeamento de dados do modelo relacional para a representação em grafos. A principal contribuição é que a escolha do modelo do grafo deve considerar o tipo de análise topológica que será realizada pelo usuário. Experimentos são apresentados e mostram resultados interessantes, incluindo heurísticas para apoiar o usuário na escolha da modelagem do grafo. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63265 2017-01-01T00:00:00Z A importância de trazer dados do modelo relacional para outros modelos e tecnologias tem sido amplamente debatidos, como por exemplo a publicação de dados como grafos. Este modelo permite executar análises topológicas, tal como ocorre nas análises de redes sociais, predição de ligações e sistemas de recomendação. Existem iniciativas para mapear de um banco de dados relacional para a representação em grafo. No entanto, eles não consideram as diferentes maneiras de gerar esses grafos, especialmente quando o objetivo é realizar análises topológicas. Este trabalho propõe heurísticas para facilitar a sistematização do mapeamento de dados do modelo relacional para a representação em grafos. A principal contribuição é que a escolha do modelo do grafo deve considerar o tipo de análise topológica que será realizada pelo usuário. Experimentos são apresentados e mostram resultados interessantes, incluindo heurísticas para apoiar o usuário na escolha da modelagem do grafo. http://sedici.unlp.edu.ar:80/handle/10915/63263 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) This work is focused on the exploration and application of entities extraction techniques for the codification and identification of geographical locations present in the geographic distribution section within botanic documents, such as the plant species manual of Costa Rica. Several technologies must be combined to achieve such objective, among them is Natural Language Processing (NLP) that helps in the extraction of entities with the usage of gazetteers. Another technology is the usage of rules (regular expressions, Deterministic Automata, context-free grammars). Additional to the identification and codification, an algorithm to bind the place names extracted to authorized sources such as gazetteer is presented. This algorithm identifies and enriches the entry text with extra information, extracted from the paragraphs where the distribution is defined in a semi unstructured text. The values of interest for this work are: world and Costa Rica distribution. After those values are identified, the information can be processed and become useful for diverse applications, such as geographic information systems. Other research projects might be interested in the results of this project. The evaluation consists in manually judging randomly selected sample of the results to establish if the algorithm yields useful data. The judgment features the evaluation of the world and Costa Rica distribution using the source context, given 3 possible values: GOOD, BAD, UNKNOWN. The ideal is to have the least BAD percentage. The algorithm is relatively good to geo-code and bind the world distribution. More work needs to be done for the Costa Rica distribution. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63263 2017-01-01T00:00:00Z This work is focused on the exploration and application of entities extraction techniques for the codification and identification of geographical locations present in the geographic distribution section within botanic documents, such as the plant species manual of Costa Rica. Several technologies must be combined to achieve such objective, among them is Natural Language Processing (NLP) that helps in the extraction of entities with the usage of gazetteers. Another technology is the usage of rules (regular expressions, Deterministic Automata, context-free grammars). Additional to the identification and codification, an algorithm to bind the place names extracted to authorized sources such as gazetteer is presented. This algorithm identifies and enriches the entry text with extra information, extracted from the paragraphs where the distribution is defined in a semi unstructured text. The values of interest for this work are: world and Costa Rica distribution. After those values are identified, the information can be processed and become useful for diverse applications, such as geographic information systems. Other research projects might be interested in the results of this project. The evaluation consists in manually judging randomly selected sample of the results to establish if the algorithm yields useful data. The judgment features the evaluation of the world and Costa Rica distribution using the source context, given 3 possible values: GOOD, BAD, UNKNOWN. The ideal is to have the least BAD percentage. The algorithm is relatively good to geo-code and bind the world distribution. More work needs to be done for the Costa Rica distribution. http://sedici.unlp.edu.ar:80/handle/10915/63262 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) Web navigation response times are affected by the heterogeneous nature of Internet: links with disparate bandwidth, servers and routers with diverse specifications and network segments with particular traffic shaping policies. Also, on a daily basis, several new information systems gets connected to the Internet consequently increasing traffic. Aiming to reduce response times, prefetching systems predicts which objects will be requested by users based on predictive models. This way, predicted objects are requested and stored in cache in advance. However, literature shows that current prefecthing systems does not contemplate how effective the predictions are. This paper presents a feedback based prefecthing system. The proposed approach considers previous predictions hits and misses to update the predictive model thus enhancing the precision of predictions. The experimental study shows an improvement on users response time obtained by the proposed system. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63262 2017-01-01T00:00:00Z Web navigation response times are affected by the heterogeneous nature of Internet: links with disparate bandwidth, servers and routers with diverse specifications and network segments with particular traffic shaping policies. Also, on a daily basis, several new information systems gets connected to the Internet consequently increasing traffic. Aiming to reduce response times, prefetching systems predicts which objects will be requested by users based on predictive models. This way, predicted objects are requested and stored in cache in advance. However, literature shows that current prefecthing systems does not contemplate how effective the predictions are. This paper presents a feedback based prefecthing system. The proposed approach considers previous predictions hits and misses to update the predictive model thus enhancing the precision of predictions. The experimental study shows an improvement on users response time obtained by the proposed system. http://sedici.unlp.edu.ar:80/handle/10915/63209 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) El agrupamiento de recursos web es una tarea significativa y difícil. En trabajos anteriores se sugirieron distintas formas de unificar fuentes de información web, tratando de solucionar inconvenientes como heterogeneidad de contenido, falta de estructura, disponibilidad, distribución, cantidad y calidad. No obstante, existen dominios más complejos que otros, donde encontrar la solución es un problema aún más engorroso, debido a la cantidad y variedad de información que se maneja y a que el conjunto de fuentes que se consultan es considerable. Un ejemplo concreto es el dominio del turismo. Si bien existen prototipos y aplicaciones comerciales que asisten al turista web, actualmente denominados sistemas recomendadores de paquetes turísticos, muchas fuentes de información se duplican y dispersan en la red. El desafío de una aplicación que brinde información concisa y unificada va más allá de ubicar dónde está esa información, procurando además saber de qué forma llegar a ella y cómo agruparla. En este trabajo se presenta una metodología para recuperar recursos web, de la cual forma parte un mecanismo para agrupar esos recursos, permitiendo el solapamiento entre grupos. Por consiguiente, con esta contribución se logra mejorar la precisión en las respuestas para un dominio establecido a priori. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63209 2017-01-01T00:00:00Z El agrupamiento de recursos web es una tarea significativa y difícil. En trabajos anteriores se sugirieron distintas formas de unificar fuentes de información web, tratando de solucionar inconvenientes como heterogeneidad de contenido, falta de estructura, disponibilidad, distribución, cantidad y calidad. No obstante, existen dominios más complejos que otros, donde encontrar la solución es un problema aún más engorroso, debido a la cantidad y variedad de información que se maneja y a que el conjunto de fuentes que se consultan es considerable. Un ejemplo concreto es el dominio del turismo. Si bien existen prototipos y aplicaciones comerciales que asisten al turista web, actualmente denominados sistemas recomendadores de paquetes turísticos, muchas fuentes de información se duplican y dispersan en la red. El desafío de una aplicación que brinde información concisa y unificada va más allá de ubicar dónde está esa información, procurando además saber de qué forma llegar a ella y cómo agruparla. En este trabajo se presenta una metodología para recuperar recursos web, de la cual forma parte un mecanismo para agrupar esos recursos, permitiendo el solapamiento entre grupos. Por consiguiente, con esta contribución se logra mejorar la precisión en las respuestas para un dominio establecido a priori. http://sedici.unlp.edu.ar:80/handle/10915/63208 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) Abstract—Social media are increasingly being used as sources in mainstream news coverage. However, since news is so rapidly updating it is very easy to fall into the trap of believing everything as truth. Spam content usually refers to the information that goes viral and skews users views on subjects. Despite recent advances in spam analysis methods, it is still a challenging task to extract accurate and useful information from tweets. This paper aims at introducing a new approach for classification of spam and non-spam tweets using Cost-Sensitive Classifier that includes Random Forest. The approach consisted of three phases: preprocessing, classification and evaluation. In the preprocessing phase, tweets were first annotated manually and then four different sets of features were extracted from them. In the classification phase, four machine learning algorithms were first cross-validated aiming at determining the best base classifier for spam detection. Then, class imbalanced problem was dealt by resampling and incorporating arbitrary misclassification costs into the learning process. In the evaluation phase, the trained algorithm was tested with unseen tweets. Experimental results showed that the proposed approach helped mitigate overfitting and reduced classification error by achieving an overall accuracy of 89.14% in training and 76.82% in testing. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63208 2017-01-01T00:00:00Z Abstract—Social media are increasingly being used as sources in mainstream news coverage. However, since news is so rapidly updating it is very easy to fall into the trap of believing everything as truth. Spam content usually refers to the information that goes viral and skews users views on subjects. Despite recent advances in spam analysis methods, it is still a challenging task to extract accurate and useful information from tweets. This paper aims at introducing a new approach for classification of spam and non-spam tweets using Cost-Sensitive Classifier that includes Random Forest. The approach consisted of three phases: preprocessing, classification and evaluation. In the preprocessing phase, tweets were first annotated manually and then four different sets of features were extracted from them. In the classification phase, four machine learning algorithms were first cross-validated aiming at determining the best base classifier for spam detection. Then, class imbalanced problem was dealt by resampling and incorporating arbitrary misclassification costs into the learning process. In the evaluation phase, the trained algorithm was tested with unseen tweets. Experimental results showed that the proposed approach helped mitigate overfitting and reduced classification error by achieving an overall accuracy of 89.14% in training and 76.82% in testing. http://sedici.unlp.edu.ar:80/handle/10915/63207 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) Author reputation is a very important variable for evaluating web comments. However, there is no formal definition for calculating its value. This paper presents an adaptation of the approach presented by Sousa (2015) for evaluating the importance of comments about products and services available online, emphasizing measures of author reputation. The implemented adaptation consists in defining six measures for authors, used as input in a Multilayer Perceptron Artificial Neural Network. On a preliminary evaluation, the Neural Network presented an accuracy of 91.01% on the author classification process. Additionally, an experiment was conduced aiming to compare both approaches, and the results show that the adapted approach had better performance in classifying the importance of comments. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63207 2017-01-01T00:00:00Z Author reputation is a very important variable for evaluating web comments. However, there is no formal definition for calculating its value. This paper presents an adaptation of the approach presented by Sousa (2015) for evaluating the importance of comments about products and services available online, emphasizing measures of author reputation. The implemented adaptation consists in defining six measures for authors, used as input in a Multilayer Perceptron Artificial Neural Network. On a preliminary evaluation, the Neural Network presented an accuracy of 91.01% on the author classification process. Additionally, an experiment was conduced aiming to compare both approaches, and the results show that the adapted approach had better performance in classifying the importance of comments. http://sedici.unlp.edu.ar:80/handle/10915/63206 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) Over the past few years, neural networks have reemerged as powerful machine-learning models, yielding state-ofthe- art results in fields such as image recognition and speech processing. More recently, neural network models started to be applied also to textual natural language signals, again with very promising results. This paper show a series of experiments with Convolutional Neural Networks for sentence-level classification tasks with different hyperparameter settings and how sensitive model performance is to changes in these configurations. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63206 2017-01-01T00:00:00Z Over the past few years, neural networks have reemerged as powerful machine-learning models, yielding state-ofthe- art results in fields such as image recognition and speech processing. More recently, neural network models started to be applied also to textual natural language signals, again with very promising results. This paper show a series of experiments with Convolutional Neural Networks for sentence-level classification tasks with different hyperparameter settings and how sensitive model performance is to changes in these configurations. http://sedici.unlp.edu.ar:80/handle/10915/63204 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) En este trabajo se describe una propuesta para la construcción rápida y robusta de sistemas de información que contengan lógicas complejas y que puedan ser desplegados en distintos ambientes y plataformas computacionales. Para ello, se hace uso de los conceptos fundamentales de base de datos relacionales, redes de petri y flujos de trabajo, así como también se toma como referencia los lenguajes de procesos de negocio, tales como BPM y BPEL. Con la finalidad de ofrecer una visión modular se discute sobre los diferentes componentes que conforman el sistema SAFET, elemento central del trabajo, subrayando los procesos de coordinación e integración entre ellos. Finalmente, se muestran dos ejemplos de uso común en internet, donde se pueden identificar de forma práctica algunos aportes de la propuesta al desarrollo de sistemas de información. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63204 2017-01-01T00:00:00Z En este trabajo se describe una propuesta para la construcción rápida y robusta de sistemas de información que contengan lógicas complejas y que puedan ser desplegados en distintos ambientes y plataformas computacionales. Para ello, se hace uso de los conceptos fundamentales de base de datos relacionales, redes de petri y flujos de trabajo, así como también se toma como referencia los lenguajes de procesos de negocio, tales como BPM y BPEL. Con la finalidad de ofrecer una visión modular se discute sobre los diferentes componentes que conforman el sistema SAFET, elemento central del trabajo, subrayando los procesos de coordinación e integración entre ellos. Finalmente, se muestran dos ejemplos de uso común en internet, donde se pueden identificar de forma práctica algunos aportes de la propuesta al desarrollo de sistemas de información. http://sedici.unlp.edu.ar:80/handle/10915/63203 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) The Processing Architecture based on Measurement Metadata (PAbMM) is a data stream management system specialized in measurement and evaluation (M&E) projects, which incorporates predictive and detective behavior on data streams. It uses a case based organizational memory for recommending courses of action in each detected online situation and previously modeled by the project definition. In this work the storm topology associated with the online processing in PAbMM is described. Additionally, a new synopses strategy for monitoring entities under analysis is presented and a new schema for training the online classifiers is introduced. This new schema allows indicating to the classifiers the problem characterization, the proposal solution and the associated indicator value (target class). A practical case associated with the weather radar of the Experimental Agricultural Station (EAS) INTA Anguil (Province of La Pampa, Argentina) is shown, indicating the advantages of this storm topology and the new schema oriented to training data set. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63203 2017-01-01T00:00:00Z The Processing Architecture based on Measurement Metadata (PAbMM) is a data stream management system specialized in measurement and evaluation (M&E) projects, which incorporates predictive and detective behavior on data streams. It uses a case based organizational memory for recommending courses of action in each detected online situation and previously modeled by the project definition. In this work the storm topology associated with the online processing in PAbMM is described. Additionally, a new synopses strategy for monitoring entities under analysis is presented and a new schema for training the online classifiers is introduced. This new schema allows indicating to the classifiers the problem characterization, the proposal solution and the associated indicator value (target class). A practical case associated with the weather radar of the Experimental Agricultural Station (EAS) INTA Anguil (Province of La Pampa, Argentina) is shown, indicating the advantages of this storm topology and the new schema oriented to training data set. http://sedici.unlp.edu.ar:80/handle/10915/63201 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) The current drug development pipelines are characterised by long processes with high attrition rates and elevated costs. More than 80% of new compounds fail in the later stages of testing due to severe side-effects caused by unknown biomolecular targets of the compounds. In this work, we present a measure that can predict shared targets for drugs in DrugBank through large scale analysis of the biomedical literature. We show that using MeSH ontology terms can accurately describe the drugs and that appropriate use of the MeSH ontological structure can determine pairwise drug similarity. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63201 2017-01-01T00:00:00Z The current drug development pipelines are characterised by long processes with high attrition rates and elevated costs. More than 80% of new compounds fail in the later stages of testing due to severe side-effects caused by unknown biomolecular targets of the compounds. In this work, we present a measure that can predict shared targets for drugs in DrugBank through large scale analysis of the biomedical literature. We show that using MeSH ontology terms can accurately describe the drugs and that appropriate use of the MeSH ontological structure can determine pairwise drug similarity. http://sedici.unlp.edu.ar:80/handle/10915/63199 Objeto de conferencia Simposio Latinoamericano de Manejo de Datos e Información (SLMDI) - JAIIO 46 (Córdoba, 2017) Multi-Agent Systems (MASs) have received much attention in recent years because of their advantages on modeling complex distributed systems, such Digital Ecosystems (DESs). Many existing modeling languages that support the design of such systems are based on ontologies to assist the representation of agents knowledge. However, in the context of DESs, there is still a need for more general conceptual models to represent the specific characteristics of DESs in terms of win-win interaction, engagement, equilibrium, and self-organization. Then, concepts such behavior, roles, rules, and environment are needed. This paper describes an ontologybased approach by proposing MAS2DES-Onto, as the conceptual model, which considers the essential static and dynamic aspects of MASs by a clear representation of their concepts and relationships to support the design and development of DESs. To validate and conduct experimental tests, we integrate MAS2DES-Onto into a framework to automatically generate MAS-based DESs. Results show the efficiency and effectiveness of our approach. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63199 2017-01-01T00:00:00Z Multi-Agent Systems (MASs) have received much attention in recent years because of their advantages on modeling complex distributed systems, such Digital Ecosystems (DESs). Many existing modeling languages that support the design of such systems are based on ontologies to assist the representation of agents knowledge. However, in the context of DESs, there is still a need for more general conceptual models to represent the specific characteristics of DESs in terms of win-win interaction, engagement, equilibrium, and self-organization. Then, concepts such behavior, roles, rules, and environment are needed. This paper describes an ontologybased approach by proposing MAS2DES-Onto, as the conceptual model, which considers the essential static and dynamic aspects of MASs by a clear representation of their concepts and relationships to support the design and development of DESs. To validate and conduct experimental tests, we integrate MAS2DES-Onto into a framework to automatically generate MAS-based DESs. Results show the efficiency and effectiveness of our approach. http://sedici.unlp.edu.ar:80/handle/10915/63179 Objeto de conferencia Simposio Argentino de GRANdes DAtos (AGRANDA) - JAIIO 46 (Córdoba, 2017) Abstract—From a network medicine perspective, diseases are caused by perturbations in the dynamics of multiple interacting genes - a disease module. A drug that is a suitable candidate for re-purposing, should affect perturbed disease modules other than the one for which it was designed. In other words, it must act on various disease modules. A systematic analysis of re purposing suitability requires deeper understanding of drug target modularity. In this paper, we present a large-scale analysis of drug-target relationships, evaluating the locality of drug targets in protein-protein interaction networks. We show that the various drugs in each category affect different regions in biological networks, and present modular features. Additionally, multiple targets associated to the same drug appear close in the interactome. Our statistical analysis of the functions of the known drug targets reveals that peripheral functions of disease modules, such as signalling, are common targets for many drugs. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63179 2017-01-01T00:00:00Z Abstract—From a network medicine perspective, diseases are caused by perturbations in the dynamics of multiple interacting genes - a disease module. A drug that is a suitable candidate for re-purposing, should affect perturbed disease modules other than the one for which it was designed. In other words, it must act on various disease modules. A systematic analysis of re purposing suitability requires deeper understanding of drug target modularity. In this paper, we present a large-scale analysis of drug-target relationships, evaluating the locality of drug targets in protein-protein interaction networks. We show that the various drugs in each category affect different regions in biological networks, and present modular features. Additionally, multiple targets associated to the same drug appear close in the interactome. Our statistical analysis of the functions of the known drug targets reveals that peripheral functions of disease modules, such as signalling, are common targets for many drugs. http://sedici.unlp.edu.ar:80/handle/10915/63178 Objeto de conferencia Simposio Argentino de GRANdes DAtos (AGRANDA) - JAIIO 46 (Córdoba, 2017) As investigações em análise automáticas de documentos vêm permitindo grandes avanços permitindo no reconhecimento de aspectos subjetivos. Dentre estes, destaca-se a classificação da polaridade do texto, ou seja, o quão negativa ou positiva são as opiniões expressadas nele. Contudo, por ser uma área ainda em desenvolvimento, métodos criados para estas análises, na maioria, são para língua inglesa, o que dificulta sua utilização em textos escritos em português. Assim, esse trabalho tem como objetivo o estudo e a aplicação de uma ferramenta que, conterá um algoritmo de classificação de sentimentos, sendo ele capaz de avaliar a polaridade de comentário extraído de uma base de dados de livros da Amazon e outra do Twitter, baseando-se em técnicas de mineração de textos. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63178 2017-01-01T00:00:00Z As investigações em análise automáticas de documentos vêm permitindo grandes avanços permitindo no reconhecimento de aspectos subjetivos. Dentre estes, destaca-se a classificação da polaridade do texto, ou seja, o quão negativa ou positiva são as opiniões expressadas nele. Contudo, por ser uma área ainda em desenvolvimento, métodos criados para estas análises, na maioria, são para língua inglesa, o que dificulta sua utilização em textos escritos em português. Assim, esse trabalho tem como objetivo o estudo e a aplicação de uma ferramenta que, conterá um algoritmo de classificação de sentimentos, sendo ele capaz de avaliar a polaridade de comentário extraído de uma base de dados de livros da Amazon e outra do Twitter, baseando-se em técnicas de mineração de textos. http://sedici.unlp.edu.ar:80/handle/10915/63177 Objeto de conferencia Simposio Argentino de GRANdes DAtos (AGRANDA) - JAIIO 46 (Córdoba, 2017) Este trabajo describe cómo se llevó adelante la implementación de herramientas de Business Intelligence en el sistema universitario nacional de Argentina. La creación del Sistema de Información Universitaria (SIU) desde 1996, para el desarrollo y la puesta en marcha de sistemas transaccionales permitió a las universidades ordenar distintos aspectos de la gestión, brindar más y mejores servicios y generar datos con mayor calidad. Una vez conseguido ese primer objetivo fue posible comenzar a trabajar en el desarrollo de modelos de análisis que permitieran transformar los datos en información valiosa para la universidad. Para hacer frente a ese nuevo desafío el SIU formó un equipo para trabajar con herramientas de inteligencia de negocios, que finalmente arribó en el desarrollo del módulo de información gerencial SIU-Wichi. Cómo se llevó a cabo esa tarea, qué herramientas se eligieron y por qué, qué papel tuvieron las universidades, los usuarios, técnicos y el SIU en este proceso, cuáles son los principales problemas que se debieron sortear y cuál es el futuro de la explotación de datos en el sistema universitario argentino son los temas más relevantes que se abordan. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63177 2017-01-01T00:00:00Z Este trabajo describe cómo se llevó adelante la implementación de herramientas de Business Intelligence en el sistema universitario nacional de Argentina. La creación del Sistema de Información Universitaria (SIU) desde 1996, para el desarrollo y la puesta en marcha de sistemas transaccionales permitió a las universidades ordenar distintos aspectos de la gestión, brindar más y mejores servicios y generar datos con mayor calidad. Una vez conseguido ese primer objetivo fue posible comenzar a trabajar en el desarrollo de modelos de análisis que permitieran transformar los datos en información valiosa para la universidad. Para hacer frente a ese nuevo desafío el SIU formó un equipo para trabajar con herramientas de inteligencia de negocios, que finalmente arribó en el desarrollo del módulo de información gerencial SIU-Wichi. Cómo se llevó a cabo esa tarea, qué herramientas se eligieron y por qué, qué papel tuvieron las universidades, los usuarios, técnicos y el SIU en este proceso, cuáles son los principales problemas que se debieron sortear y cuál es el futuro de la explotación de datos en el sistema universitario argentino son los temas más relevantes que se abordan. http://sedici.unlp.edu.ar:80/handle/10915/63175 Objeto de conferencia Simposio Argentino de GRANdes DAtos (AGRANDA) - JAIIO 46 (Córdoba, 2017) Fraud in the mobile advertising world is a topic gaining momentum recently. Different reports agree that invalid traffc is generating losses in the order of billions of dollars and that there is a signi cant amount of fraud with ongoing efforts against it. Here at Jampp1 we use an automated fraud detection algorithm for a recent type of mobile advertising fraud, which can be referred to as click-spamming, click-injection or mobile-hijacking. We propose a metric to measure suspicious installs, and use a heuristic to compare the ts of theoretical distributions to this metric. This allows us to derive a threshold for suspicious installs. Our metric is based on the time-delta distributions, which amounts to the time it takes from a click in an ad to be converted into an install. The model is currently in use with satisfactory results. To the best of our knowledge, this is the rst algorithm in production used to tackle this speci c kind of fraud. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63175 2017-01-01T00:00:00Z Fraud in the mobile advertising world is a topic gaining momentum recently. Different reports agree that invalid traffc is generating losses in the order of billions of dollars and that there is a signi cant amount of fraud with ongoing efforts against it. Here at Jampp1 we use an automated fraud detection algorithm for a recent type of mobile advertising fraud, which can be referred to as click-spamming, click-injection or mobile-hijacking. We propose a metric to measure suspicious installs, and use a heuristic to compare the ts of theoretical distributions to this metric. This allows us to derive a threshold for suspicious installs. Our metric is based on the time-delta distributions, which amounts to the time it takes from a click in an ad to be converted into an install. The model is currently in use with satisfactory results. To the best of our knowledge, this is the rst algorithm in production used to tackle this speci c kind of fraud. http://sedici.unlp.edu.ar:80/handle/10915/63174 Objeto de conferencia Simposio Argentino de GRANdes DAtos (AGRANDA) - JAIIO 46 (Córdoba, 2017) Abstract—Patterns of mobile phone communications, coupled with the information of the social network graph and financial behavior, allow us to make inferences of users’ socio-economic attributes such as their income level. We present here several methods to extract features from mobile phone usage (calls and messages), and compare different combinations of supervised machine learning techniques and sets of features used as input for the inference of users’ income. Our experimental results show that the Bayesian method based on the communication graph outperforms standard machine learning algorithms using nodebased features. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63174 2017-01-01T00:00:00Z Abstract—Patterns of mobile phone communications, coupled with the information of the social network graph and financial behavior, allow us to make inferences of users’ socio-economic attributes such as their income level. We present here several methods to extract features from mobile phone usage (calls and messages), and compare different combinations of supervised machine learning techniques and sets of features used as input for the inference of users’ income. Our experimental results show that the Bayesian method based on the communication graph outperforms standard machine learning algorithms using nodebased features. http://sedici.unlp.edu.ar:80/handle/10915/63173 Objeto de conferencia Simposio Argentino de GRANdes DAtos (AGRANDA) - JAIIO 46 (Córdoba, 2017) La administración de la hacienda en un mercado concentrador como el Mercado de Liniers requiere de una estimación diaria de la cantidad de cabezas disponibles para su ingreso. La confluencia de intereses de los actores participantes en este mercado determina que el ingreso total de hacienda sea una de las variables más consultadas diariamente. Predecir esta variable es importante tanto para los diferentes actores de la cadena de comercialización de hacienda, organismos gubernamentales de control, como también para otros actores de los negocios agropecuarios en general. En este trabajo se presenta un algoritmo de estimación del ingreso de hacienda para intentar pronosticar el ingreso total de animales al Mercado de Liniers en jornadas futuras. A partir del procesamiento de una dataset específico y su posterior enriquecimiento, se presenta un algoritmo supervisado de clasificación predictivo basado en árboles de decisión. Esta nueva herramienta intenta ser un aporte adicional a la transparencia y a la previsibilidad de la operatoria de compraventa de hacienda en el mercado formador de precios de referencia en el sector agropecuario. Sun, 01 Jan 2017 00:00:00 GMT http://sedici.unlp.edu.ar:80/handle/10915/63173 2017-01-01T00:00:00Z La administración de la hacienda en un mercado concentrador como el Mercado de Liniers requiere de una estimación diaria de la cantidad de cabezas disponibles para su ingreso. La confluencia de intereses de los actores participantes en este mercado determina que el ingreso total de hacienda sea una de las variables más consultadas diariamente. Predecir esta variable es importante tanto para los diferentes actores de la cadena de comercialización de hacienda, organismos gubernamentales de control, como también para otros actores de los negocios agropecuarios en general. En este trabajo se presenta un algoritmo de estimación del ingreso de hacienda para intentar pronosticar el ingreso total de animales al Mercado de Liniers en jornadas futuras. A partir del procesamiento de una dataset específico y su posterior enriquecimiento, se presenta un algoritmo supervisado de clasificación predictivo basado en árboles de decisión. Esta nueva herramienta intenta ser un aporte adicional a la transparencia y a la previsibilidad de la operatoria de compraventa de hacienda en el mercado formador de precios de referencia en el sector agropecuario.