Insights: Knowledge Mining software that extracts knowledge from data. Gather. 
    		Mine. Extract. Predict. Identify. Simulate. Discover. Associate. Easily and Reliably.
    		Ultra-fast, parallel, self-organizing, high-dimensional modeling of real-world processes and complex systems.
Selected Publications.

Self-organizing Modeling applied in different domains.

This page provides selected papers related to application of modeling technologies related to KnowledgeMiner Insights.

Prediction of Wastewater Pre-precipitation Variables using Self-Organizing Networks
P. Isaias, S. Nilsson, A. Stathaki and R. E. King

presented at the 13th IEEE Mediterranean Control Conference - MED'05, Limassol, Cyprus, June 2005


This paper describes the derivation and design of an array of self-organizing networks trained by inductive learning for one step ahead prediction of the outputs of the pre-precipitation stage of a wastewater treatment plant with a view to model predictive control of the stage.
Full Paper: (116k - PDF, Proceedings 13th IEEE MED'05 Conference, 2005

A GMDH Approach to Modelling Gibbsite Solubility in Bayer Process Liquors
Frederick R. Bennett , Peter Crew, Jennifer K. Muller

CSIRO Minerals, Western Australia
Received: 4 May 2003 / Accepted: 29 August 2003 / Published: 20 February 2004


The most widely employed industrial process for producing alumina (Bayer process) involves the dissolution of available aluminium hydroxide minerals present in raw bauxite into high temperature sodium hydroxide solutions. On cooling of the solution, or liquor in the industrial vernacular, Al is precipitated from solution in the form of gibbsite (Al(OH)3). In order to optimise the process, a detailed knowledge of factors influencing gibbsite solubility is required, a problem that is confounded by the presence of liquor impurities. In this paper, the use of the Group Method of Data Handling (GMDH) polynomial neural network for developing a gibbsite equilibrium solubility model for Bayer process liquors is discussed. The resulting predictive model appears to correctly incorporate the effects of liquor impurities and is found to offer a level of performance comparable to the most sophisticated phenomenological model presented to date.
Full Paper: (156k - PDF, International Journal of Molecular Sciences, 2004, 5, 101-109 ©MDPI)

Group method of data handling (GMDH) in modelling of growth dynamics of trees irrigated with wastewater
Ioannis Kalavrouziotis (1) , Volodymyr Stepashko (2), Vladimir Vissikirsky (2), Panagiotis Drakatos (3)

(1) Department of Environmental and Natural Resources Management , University of Ioannina, Greece
(2) Department for Information Technologies of Inductive Modeling, the International UNESCO Center of Information Technologies and Systems of the National Academy of Sciences of Ukraine
(3) Department of Mechanical and Aeronautics Engineering, Laboratory of Special Mechanical Engineering, Rion, University of Patras, Greece


This paper describes the formulation and analysis of growth dynamics models for trees irrigated with wastewater. The models can be used to obtain the characteristics of a species, depending on treatment conditions and climate factors, at all stages of growth. The experiments were carried out at the University of Patras to identify the characteristics (height rate and mortality) of the forest tree Pinus brutia cultivated under different treatment conditions. The growth dynamics models are designed on the basis of the group method of data handling. This principle generates sets of estimation models with different complexity and accuracy. By analysing their structures, qualitative features of the models may be assessed, and general linear models for different treatment cases compiled.
Link to Full Paper: (International Journal of Environment and Pollution (IJEP), Vol. 21, No. 4, 2004 ©InderScience Publishers)

Forecasting Commodity Prices for Predictive Decision Support Systems
Tamer Shahwan (1), Frank Lemke (2)

(1) Humboldt-Universitaet zu Berlin, School of Business and Economics, Institute of Banking, Stock Exchanges and Insurance, Germany
(2) KnowledgeMiner Software, Germany


The builders of traditional decision support systems have regularly used game theory and operations research to build intelligent decision support systems. However, we have seen a rapid acceptance of new technology like neural networks and data mining which implement state-of-the art decision support system based on knowledge management to solve a wide range of business problems like classification and forecasting. Thus, we will illustrate our claims through the investigation of the forecasting performance of two time series forecasting techniques, namely elman neural network and self-organizing data mining against the autoregressive integrated moving average (ARIMA) model and futures prices as benchmarks. As an attempt to improve the accuracy of elman neural networks, the global search capability of Genetic Algorithms will be used to determine the optimal architecture of elman neural network. Real data sets of commodity prices are used to examine the forecasting accuracy of the proposed models.
Full Paper: (396k - PDF, Proceedings of EFITA and WCCA 2005, joint congress on on information technology in agriculture (25-28 July 2005), Vila Real, Portugal)

Data-driven Modeling and Prediction of Acute Toxicity of Pesticide Residues
F.Lemke (1), E. Benfenati (2), J.-A. Mueller (1)

(1) KnowledgeMiner Software, Germany
(2) Istituto di Ricerche Farmacologiche "Mario Negri", Milan, Italy


This paper outlines and implements a concept for developing alternative tools for toxicity modeling and prediction of chemical compounds to be used for evaluation and authorization purposes of public regulatory bodies to help minimizing animal tests, costs, and time associated with registration and risk assessment processes. Starting from a general problem description we address and introduce concepts of multileveled self-organization for high-dimensional modeling, model validation, model combining, and decision support within the frame of a knowledge discovery from noisy data.
Full Paper: (1.2m - PDF, SIGKDD Explorations Special Issue, 8(2006)1, pp. 71-79)