WekaBioSimilarity --- extending Weka with resemblance measures.

Autor: Heras Vicente, JonathanDomínguez Pérez, CésarPascual Martínez-Losa, María VicoMata Sotés, Eloy Javier

Tipo de documento: Capítulo de libro

Libro: Lecture Notes in Artificial Intelligence. Proceedings of CAEPIA 2016.

Año: 2016  ISBN: 978-3-319-44635-6  Páginas: 89-98

Texto completo open access 

Resumen: The classification of organisms is a daily-basis task in biology as well as other contexts. This process is usually carried out by comparing a set of descriptors associated with each object. However, general-purpose statistical packages offer a limited number of methods to perform such a comparison, and specific tools are required for each concrete problem. Weka is a freely-available framework that supports both supervised and unsupervised machine-learning algorithms. Here, we present WekaBioSimilarity, an extension of Weka implementing several resemblance measures to compare different kinds of descriptors. Namely, WekaBioSimilarity works with binary, multi-value, string, numerical, and heterogeneous data. WekaBioSimilarity, together with Weka, offers the functionality to classify objects using different resemblance measures, and clustering and classification algorithms. The combination of these two systems can be used as a standalone application or can be incorporated in the workflow of other software systems that require a classification process. WekaBioSimilarity is available at http://wekabiosimilarity.sourceforge.net.