Keywords
knowledge acquisition and management, data mining, machine learning, environmental databases, statistical modelling, rules induction, wastewater treatment plant
Start Date
1-7-2004 12:00 AM
Abstract
In this work, last results of the research project “Development of an Intelligent Data Analysis System for Knowledge Management in Environmental Data Bases (DB)” are presented. The project is focussed on the design and development of a prototype for Knowledge Discovery (KD) and intelligent data analysis, and specially oriented to environmental DB. It is remarkable the high quantity of information and knowledge patterns that are implicit in large DB coming from environmental domains. In this project, several environmental DB such as meteorological phenomena, wastewater treatment plants (WWTP), or environmental emergencies were used for testing. KD is a prior and mandatory step to get reliable Intelligent Environmental Decision Support Systems. Although in the literature other KD tools exists (WEKA, Intelligent Miner…) none of them integrate, like GESCONDA, statistical and AI methods, the possibility of explicit management of the produced knowledge in Knowledge Bases (KB) (in the classical AI sense), mixed techniques that can cooperate among them to discover and extract the knowledge contained in data, dynamical data analysis… in a single tool, allowing interaction among all the methods. The purpose of the paper is to present the final architecture of GESCONDA, as well as some of the methods incorporated in last phases. Later, an application to discover knowledge patterns from an environmental DB (a WWTP) is detailed. The DB has been mined using several methods available in GESCONDA. First of all, statistical filtering approaches were applied for data preparation. Afterwards, a hybrid clustering technique (clustering based on rules) was used to discover the structure of the target phenomenon. Finally, clustering results were used as input for rule induction making new knowledge explicit. Results and feedback from validation steps show that the tool seems to be useful and efficient for KD.
Knowledge Discovery in Environmental Data Bases using GESCONDA
In this work, last results of the research project “Development of an Intelligent Data Analysis System for Knowledge Management in Environmental Data Bases (DB)” are presented. The project is focussed on the design and development of a prototype for Knowledge Discovery (KD) and intelligent data analysis, and specially oriented to environmental DB. It is remarkable the high quantity of information and knowledge patterns that are implicit in large DB coming from environmental domains. In this project, several environmental DB such as meteorological phenomena, wastewater treatment plants (WWTP), or environmental emergencies were used for testing. KD is a prior and mandatory step to get reliable Intelligent Environmental Decision Support Systems. Although in the literature other KD tools exists (WEKA, Intelligent Miner…) none of them integrate, like GESCONDA, statistical and AI methods, the possibility of explicit management of the produced knowledge in Knowledge Bases (KB) (in the classical AI sense), mixed techniques that can cooperate among them to discover and extract the knowledge contained in data, dynamical data analysis… in a single tool, allowing interaction among all the methods. The purpose of the paper is to present the final architecture of GESCONDA, as well as some of the methods incorporated in last phases. Later, an application to discover knowledge patterns from an environmental DB (a WWTP) is detailed. The DB has been mined using several methods available in GESCONDA. First of all, statistical filtering approaches were applied for data preparation. Afterwards, a hybrid clustering technique (clustering based on rules) was used to discover the structure of the target phenomenon. Finally, clustering results were used as input for rule induction making new knowledge explicit. Results and feedback from validation steps show that the tool seems to be useful and efficient for KD.
