, ,


ISSN: 0368-492X

Article publication date: 1 June 2004



Zwick, M., Shu, G. and Lin, Y. (2004), "Editorial", Kybernetes, Vol. 33 No. 5/6. https://doi.org/10.1108/k.2004.06733eaa.002



Emerald Group Publishing Limited

Copyright © 2004, Emerald Group Publishing Limited


Reconstructability analysis (RA) dates back to the pioneering work of Ashby in the mid-1960s. In the 1970s and 1980s, RA was the subject of very active research in the systems community. It receded for a time as a focus of activity, but the special issue of the International Journal of General Systems (IJGS) in 1996 on the “General Systems Problem Solver” and the special IJGS issue in 2000 on “Reconstructability Analysis in China” marked the renewal of interest in this area. The current volume is part of this resurgence of activity. It collects together papers from the group at Portland State University (Portland, Oregon, USA), from the Chinese RA workers, and from other investigators. It is evident from these papers that RA continues to be a productive research area, that it has considerable value for data analysis and data-mining, and that this value has yet to be fully explored and exploited.

This special issue starts with an overview of RA theory and methodology by Professor Martin Zwick. This overview presents the basic ideas of probabilistic (“information-theoretic”) and crisp-possibilistic (“set-theoretic”) RA, illustrating the first using health care data, and the second using mappings of elementary cellular automata. In the paper, “A comparison of modified reconstructability analysis and Ashenhurst-Curtis decomposition of Boolean functions”, Anas Al-Rabadi, Marek Perkowski and Martin Zwick present an enhanced crisp possibilistic RA and shows that it yields better decompositions for simple binary functions than a standard logic decomposition procedure. In the paper, “Modified reconstructability analysis for many-valued logic functions and relations, Anas Al-Rabadi and Martin Zwick extends the method presented in the previous paper to multi-valued relations and functions, and, in “Reversible modified reconstructability analysis of Boolean circuits and its quantum computation”, they show how it can be used in reversible and quantum computation.

In “Multi-level decomposition of probabilistic relations”, Stanislaw Grygiel, Martin Zwick and Marek Perkowski demonstrate how multi-level latent variable methods, which are common in machine learning and logic design but uncommon in the RA literature, can be applied to probabilistic RA. Susanne Hoeppner and Gary Shaffer, in “The K-systems glitch: granulation of predictor variables”, apply k-systems analysis to ecological modeling, demonstrating the power of a small number of factors to capture the essential behavior of the system and making apparent the critical nature of the granulation (binning) procedure used to discretize the systems function. In the next paper, “Directed extended dependency analysis for data mining”, Thaddeus Shannon and Marin Zwick report an improved implementation of the heuristic techniques proposed by Conant for loopless models, and apply these techniques to time-series analysis and pattern recognition.

In the paper, “Instant modeling and data-knowledge processing by reconstructability analysis”, Professor Guangfu Shu introduces RA modeling with multi-variety information and knowledge after reviewing data-driving factor RA modeling and the leveled variable factor RA modeling generation process. In the tenth article, “Application of reconstructability analysis in system structure”, Pengtao Wang and Changyun Yu establish a factor RA forecasting and talent quality model for development research on talent resource. Kenneth Willett and Martin Zwick, in “A software architecture for reconstructability analysis”, describe the architecture of the RA software package named OCCAM (“organizational complexity, computation, and modeling”); this package is user-friendly and web accessible, and is the primary research and applications platform for the Portland group.

W. Yao, C. Essex, P. Yu and M. Davison, “Forecast entropy”, apply an entropy measure to the analysis of chaotic time-series data, and show that this measure captures central dimension and delay parameters of the underlying attractor. In the 13th paper, “The forecast model of system reconstructability analysis”, Professors Changyun Yu and Pengtao Wang analyze the use of factor RA and the elastic coefficient forecast method in talent forecasting. They use this approach to study talent structure, demand and quality of trained human resources in the electronic industry. Zhihong Zhang, Pengtao Wang, Huaqing Liu, and Guangfu Shu, in “Construction of main sequence of gene based on 'method of factor reconstructability analysis'”, introduce factor RA to the study of gene sequences in the hope that this method can help us decipher, analyse and understand genomic information.

In the article, “Reconstructability analysis with Fourier transform”, Martin Zwick develops an approach to probabilistic RA utilizing Fourier transforms, which bypasses the need for iterative methods for models with loops, and requires computation which scales only with the data instead of the state space. Martin Zwick and Michael Johnson in the paper “State-based reconstructability analysis” shows that the state-based modeling idea originally introduced by Bush Jones in his “k-systems analysis” can be separated from its initial use for function approximation and integrated fully with probabilistic and statistical RA. In the final paper, Martin Zwick and Stephen Shervais demonstrate the use of RA as a preprocessor for genetic algorithms to determine the optimal order for the variables on the GA genome; this utilization of RA calls to mind other uses of RA as a preprocessor for neural nets.

As the guest editors of this special issue, we would like to express our gratitude to Professor Brian H. Rudall, Editor of the prestigious Kybernetes: The International Journal of Systems and Cybernetics, for providing this valuable opportunity and forum for us to publish these articles together as a special issue.

Martin Zwick Systems Science PhD Program, Portland State University, Portland, OR, USAE-mail: zwick@pdx.edu

Guangfu ShuInstitute of Systems Science, Academy of Mathematical and Systems Sciences, Chinese Academy of Sciences, Beijing, People's Republic of ChinaE-mail: guangfushu@mx.amss.ac.cn

Yi LinInternational Institute for General Systems Studies, Inc., 23 Kings Lane, Grove City, PA, USAE-mail: Jeffrey.forrest@sru.edu

Related articles