Development of an autonomous distributed multiagent monitoring system for the automatic classification of end users

  • Mohammed Mhereeg

    Student thesis: Doctoral Thesis


    The purpose of this study is to investigate the feasibility of constructing a software Multi-Agent based monitoring and classification system and utilizing it to provide an automated and accurate classification for end users developing applications in the spreadsheet domain. Resulting in, is the creation of the Multi-Agent Classification System (MACS). The Microsoft‘s .NET Windows Service based agents were utilized to develop the Monitoring Agents of MACS. These agents function autonomously to provide continuous and periodic monitoring of spreadsheet workbooks by content. .NET Windows Communication Foundation (WCF) Services technology was used together with the Service Oriented Architecture (SOA) approach for the distribution of the agents over the World Wide Web in order to satisfy the monitoring and classification of the multiple developer aspect. The Prometheus agent oriented design methodology and its accompanying Prometheus Design Tool (PDT) was employed for specifying and designing the agents of MACS, and Visual Studio.NET 2008 for creating the agency using visual C# programming language. MACS was evaluated against classification criteria from the literature with the support of using real-time data collected from a target group of excel spreadsheet developers over a network. The Monitoring Agents were configured to execute automatically, without any user intervention as windows service processes in the .NET web server application of the system. These distributed agents listen to and read the contents of excel spreadsheets development activities in terms of file and author properties, function and formulas used, and Visual Basic for Application (VBA) macro code constructs. Data gathered by the Monitoring Agents from various resources over a period of time was collected and filtered by a Database Updater Agent residing in the .NET client application of the system. This agent then transfers and stores the data in Oracle server database via Oracle stored procedures for further processing that leads to the classification of the end user developers. Oracle data mining classification algorithms: Naive Bayes, Adaptive Naive Bayes, Decision Trees, and Support Vector Machine were utilized to analyse the results from the data gathering process in order to automate the classification of excel spreadsheet developers. The accuracy of the predictions achieved by the models was compared. The results of the comparison showed that Naive Bayes classifier achieved the best results with accuracy of 0.978. Therefore, the MACS can be utilized to provide a Multi-Agent based automated classification solution to spreadsheet developers with a high degree of accuracy.
    Date of Award2011
    Original languageEnglish

    Cite this