Web Components for Development of Computational Methods: Example with Fuzzy Logic Rules

Abstract

Computational methods (either statistical or non-statistical) for experimental design, data analysis and optimisation are needed to provide information which can be used by a variety of decision makers and for a variety of purposes in the natural sciences research. It is crucial for scientists to keep the pace with evolving computational approaches, to comply with specific needs emerging from applied research. Software tools are desirable not only providing easy access to a large variety of data processing methods but also granting for continually updating, adding more solutions as well as keeping all the parts up to date as computing environments change and novel computational methods evolve. Based on that, the Data Analysis ¿N¿ Assessment (DANA) project was approached and meant as platform-independent, integrated repository of statistical and soft-computing procedures for use in applied research. It consists of a core engine and a set of plug-ins sharing an array of functionalities. The core engine is written in Java and integrates a set of libraries (e.g. from the R language for statistical computing, http://www.r-project.org) accessible via web services. This is achieved through usage of the object-oriented paradigm and high-level utility functions, hiding much of the complexity of the underlying libraries. The core engine handles data logging, event handling, information and synchronisation between different libraries. All classes implement a defined interface that makes it possible to create derivative works by third parties. A Graphical Users¿ Interface is dynamically generated via PHP through a set of configuration XML files. The XML Schema Definition below undergoes validation. This paper illustrates the features of DANA through the components used for computation of fuzzy-logic based rules.