No abstract

The aim of the present work is to define a theoretical framework which facilitates the analysis of the structure of a network at an individual level. To this end, we propose the integration of co-word analysis together with the analysis of social networks. As a result of this study, a series of diagrams called "structural diagrams" has been obtained. Thanks to these diagrams a pattern can be assigned to each node of the network, and a network can be classified into a set of typologies. The information provided by this theoretical framework will allow a deeper understanding of the dynamics of systems, modeled in the form of networks. In this context, structural diagrams technique improves strategies of the visual exploration of the networks as well as to orientate the definition of those procedures which enable the transformation of one typology of network into another. As a specific example of a real application of this theoretical framework, the social network of the Journal of Software Engineering and Databases (JISBD) scientific community has been analyzed, based on its co-authorship networks.

No abstract

A probabilistic interpretation for the output obtained from a tri-class Support Vector Machine into a multi-classification problem is presented in this paper. Probabilistic outputs are defined when solving a multi-class problem by using an ensemble architecture with tri-class learning machines working in parallel. This architecture enables the definition of an 'interpretation' mapping which works on signed and probabilistic outputs providing more control to the user on the classification problem.

Based on an interval distance, three functions are given in order to quantify similarities between one-dimensional data sets by using first-order statistics. The Glass Identification Database is used to illustrate how to analyse a data set prior to its classification and/or to exclude dimensions. Furthermore, a non-parametric hypothesis test is designed to show how these similarity measures, based on random samples from two populations, can be used to decide whether these populations are identical. Two comparative analyses are also carried out with a parametric test and a non-parametric test. This new non-parametric test performs reasonably well in comparison with classic tests.Key words: Data mining, Interval distance, Kernel methods, Non-parametric tests. ResumenBasadas en una distancia intervalar, se dan tres funciones para cuantificar similaridades entre conjuntos de datos unidimensionales mediante el uso de estadísticos de primer orden. Se usa la base de datos Glass Identification para ilustrar cómo esas medidas de similaridad se pueden usar para analizar un conjunto de datos antes de su clasificación y/o para excluir dimensiones. Además, se diseña un test de hipótesis no parámetrico para mostrar cómo similaridad, basadas en muestras aleatorias de dos poblaciones, se pueden usar para decidir si esas poblaciones son idénticas. También se realizan dos análisis comparativos con un test paramétrico y un test no paramétrico. Este nuevo test se comporta razonablemente bien en comparación con test clásicos.Palabras clave: distancia entre intervalos, métodos del núcleo, minería de datos, tests no paramétricos.a Senior lecturer.