Empowering Education Managers in Schools via a Multi-Agent System

With the recent advances in emerging technology, the study of intelligent agents has become one of the most important fields in education. The voluminous information available in the education management field has given rise to the exploration of agent technology to analyse data and make critical business decisions. In this paper, we propose a framework of an intelligent multi-agent based information retrieval for education management. This framework consists of two main parts – the multi-agent education management system and an ontology model. The proposed framework was implemented using the Jade Multi-Agent System and the ontology was designed using Protégé. In general, the proposed system helps school administrators to search for information precisely and rapidly. The system searches for relevant documents from various databases, parses and presents them in an XML format. This will free education administrators from relatively tedious tasks to focus more on decision-making processes.. INTRODUCTION With increasing regulatory pressure and public demands, education managers are seeking new ways to enhance effectiveness and efficiency in schools. The voluminous information available in the education management field has given rise to the exploration of agent technology to analyse data and make critical business decisions. School administrators, such as principals and senior assistants for curriculum, co-curriculum and examination secretaries, are always given many administrative tasks besides their normal teaching loads. Such work is inclusive of class management, examination enrolment, sports management, education related public events and attending numerous meetings at school, district and state levels. At each level, they are required to collect vast amounts of information on a continuous basis to be transmitted into computer systems. Although schools are moving towards a paperless environment capable of speedy submissions of information, the process of collecting, sorting and keying in data continues to be done in a paperbased administrative style. This task became more daunting as there are multiple formats for the input and output of data required for various uses and at different departmental levels. Furthermore, there is frequently a two-way document flow bottleneck between school administrators and district or state level administrators that hampers efforts to facilitate information flow and decision making. This paper is organised as follows: Section Two discusses the background and motivation issues. Section Three introduces the multi-agent architecture for the education management system (EMS). Section Four discusses the EMS ontology model. In Section Five, the overall application of the EMS for education management is discussed. Finally, Section Six contains concluding remarks. BACKGROUND AND MOTIVATION Recent studies have shown that teachers in some schools in Malaysia have been burdened with a heavy teaching and school administration workload. Statistics indicate that these teachers spend 74 hours a week on their job; 68.1 percent of this time is used to carry out curriculum-related duties, 12.7 percent co-curricular activities and 8.9 percent attending to student needs (Bernama 2005). MOJIT Empowering Education Managers in Schools via Multiagent 18 School administration is more than just handling academic matters that involve teachers and students. Modern education requires a conducive and well-equipped learning environment to operate in with a balanced growth between academic and non-academic matters. In a dynamic school environment, data come from various formats ranging from the multiple formats of paperbased reports, electronic data and video, audio and graphic materials, to name a few. If required data are supplied in a uniform format, the administrators can spend more time on decisionmaking processes that will result in quality decisions. However, this scenario rarely happens. The process of collecting and formatting data takes a great amount of time and effort. Sometimes, it has to be done manually, with errors and mistakes often occurring and inaccurate data being sent to higher administrative levels. Most schools are unable to shoulder the expensive costs of correcting these mistakes. Under certain circumstances, therefore, they can be left with no choice but to use “unclean data”. The education department continuously needs data and information from schools to make decisions. Sometimes, required information is submitted late and the information can no longer be used by the decision makers at that point of time. Education managers may be using different types of information formats in decision making and the formats and quality of available data sometimes become questionable and may jeopardise the decision-making process. Again at this level, similar repetitive processes may be utilised to sort and format information. A multi-agent system is one that consists of a number of agents which interact with one another (Wooldridge, 2002). These agents are autonomous and in control of what they do, what they need to do next and how/when to communicate with other agents (Subarmaniam & Vijanth, 2005). An agent is an executing programme that can migrate during execution from machine to machine in a heterogeneous network. In other words, it can suspend its action, migrate to another machine and then resume execution on the new machine from the part it left off (Brewington et al., 1999). On each machine, the agent interacts with other agents and other resources to accomplish its task. By moving to the location of an information resource, the agent can search the resource locally, eliminating the transfer of intermediate results across the network and reducing end-to-end latency. Even if the network link breaks down suddenly at that time, an agent can continue executing, thus making it very much desired in distributed information-retrieval applications (Wooldridge, 2002). More importantly, an agent can choose different migration techniques depending on its task and the current status of the network, and then change its strategies in line with the network environment. Complicated and efficient behaviours can be achieved with minimal coding. Various agent-based information retrieval systems have been developed to solve voluminous information management problems. CALVIN is a multi-agent based personal information retrieval system which observes users while they are accessing and searching for documents. This system offers an information environment with a standard interface (Bauer & Leake, 2002). A high level of abstractions of areas of the information retrieval task is implemented independently by each of CALVIN’s agents and in doing so, combines information from different sources. CALVIN gives the illusion that it is a single agent to its users but it is actually a multi-agent system hidden behind the interface. According to its creators, this is what enables the system architecture to be flexible and distributed. Soh (2004) discusses a multi-agent information retrieval system and the ways the agents within the information retrieval system handle queries and perform information retrieval. He also briefly discusses the distributed ontology. It contains nine modules (interface, query processor, action planner, collaboration manager, neighbourhood profiler, query composer, activity monitor, thread manager and negotiation manager) which are implemented in the C++ language. Through socket connections, each agent receives user queries from a software user and communicates with other MOJIT Empowering Education Managers in Schools via Multiagent 19 agents through a central relay server module. Each agent generates and dynamically maintains its neighbourhood profile during runtime (Soh, 2004). Reiterer and Mann (2000) describe INSYDERI, their visual information seeking system which chooses the presentations for the visualisations of text documents and incorporate them in a new style. Reiterer and Mann’s (2000) choice-presented visualisations are derived from their aim to find communicative visualisations and at the same time, not forgetting their intended user groups, their usual activities, technical surroundings and the type of data to be presented. They created INSYDER1 with the aim of “...the intelligent combination of the selected visualisation supporting different views on the retrieved document set and the documents itself...” (Reiterer and Mann, 2000). In addition, although they may not be focusing largely on the information retrieval aspect, they aim to provide a few additional features not common to the usual information retrieval systems. In this paper, we propose a framework of an intelligent multi-agent based information retrieval for education management. This framework consists of two main parts – the multi-agent based education management system (EMS) and an ontology model. The proposed framework was implemented using the Jade Multi-Agent System and the ontology was designed using Protégé. In general, the system helps school administrators to search for information precisely and rapidly. It searches for the relevant documents from various databases, parses and presents them in the XML format. Administrators will then be able to focus more on decision-making processes than on document searching and sorting activities. THE MULTI-AGENT ARCHITECTURE FOR THE EDUCATION MANAGEMENT SYSTEM The use of multi-agents provides a dynamic and robust environment within which the final output, an information retrieval EMS system, has been produced. A multi-agent system, being robust and reusable, may be easily upgraded as time goes by to provide increased performance. The XML format, because of its flexibility and the fact that it is a relatively new idea, will probably continue to be a powerful and popular tool to display documents in a Web-interface for some time. Moreover, as the research study objective was to display various forms of documents after searching a database, we proposed that the different types of documents would be retrieved and parsed in a single format. Thus, XML would make a successful candidate in assisting us in this task (Walsh, 1998). The fact that we would be able to specify our own semantics and tag set made XML a flexible tool with which we would be able to transform all retrieved documents (of various formats, e.g., texts, spreadsheets, etc.) into a standard format to be displayed to the user. The system architecture is elaborate with six types of agents, each with a specialised function: main, ontology, dispatcher, searcher, broker, saver and presenter agents. Figure 1 depicts the system architecture. The system will first initialise the main manager, working at a virtual machine. Once the main manager has started, it will start up the EMS ontology by invoking the ontology manager. which then activates the user interface agent and displays the interface to the user. The user needs to input its required parameters and search criteria or keyword. Then this input will be passed as parameters by the user interface agent to the ontology manager. MOJIT Empowering Education Managers in Schools via Multiagent 20 Figure 1: The Multi-Agent Architecture for the Education Management System The ontology manager parses the parameters to create a parameter resource description framework (RDF) which will be used as a configuration and working parameter in EMS and be sent back to the main manager. Using the parameter RDF, the main manager will invoke the dispatcher manager who in turn will invoke an appropriate dispatcher agent and pass the parameter RDF over to a searcher manager at various platforms. The searcher manager will invoke the search agent which starts searching external sources, such as files and databases, with the searching criteria specified in the parameter RDF. Search results will be returned by the search agent, and will go through the XML parsing process. The searcher manager will send the result XML to the main manager which in turn invokes the saver manager to store it. The saver manager triggers its saver agent to do this via the internal database, for record keeping and future use. When this is accomplished, the saver agent will notify the saver manager, and a message will be sent to the main manager, notifying it of the status of the storing process. The main manager then starts the presentation application by invoking the presenter manager, passing on presentation parameters in a message to present the results. The presenter manager will send an upload agent to retrieve the result XML, based on the parameters within the message sent by the main manager. Finally, the retrieved XML will be converted into XSLT by the presenter agent and displayed to the user. As shown in Figure 1, the agents in these containers are the following: the saver container, the saver manager, the saver agent, the presenter container, the presenter manager, the upload agent and the presenter agent. The ontology, together with the individual’s set of concepts, constitutes a knowledge base. An intelligent agent is a system that can perform a task based on intelligence learned or rules provided. It is able to independently evaluate choices without human interaction. When several agents are put together, they form an agency, capable of a combined range of actions. The agency picks the right agent needed to perform a specific task, using the network itself to do the processing. In other words, the agent represents the user to select and complete a required task through ruled-based criteria while using and interacting with other programmes and data. MOJIT Empowering Education Managers in Schools via Multiagent 21 THE EMS ONTOLOGY MODEL There are several definitions for ontology. Ontology has evolved its philosophical meaning to suit modern time and technology. Generally speaking, ontology means a formal, explicit specification of a shared conceptualisation (Gruber, 1993). Elsewhere, (Fensel, 2001) described ontology using four different concepts, which are conceptualisation, which means an abstract model of a phenomenon, formal, which is described by a precise mathematical description, explicit, the relationship of precision concepts and shared which means the existence of an agreement between ontology users. According to Fonseca & Egenhofer (2000), ontology refers to a theory which uses a particular set of words to explain entities, classes, properties and related functions from a certain point of view. Ontology is essential for sharing data and for interoperability between agencies. It is an annotated taxonomy of a specific component of the real world. For example, ontology about the education field would contain information about curricula, subjects, examinations, evaluation criteria and so on. In EMS ontology, education is deployed as a main domain. Both curricula and co-curricula are selected as classes because they are the two core components in the education domain. As a subclass, syllabi, subjects and evaluation methodology are designed for the curricula class while the subclass for co-curricula are the academic society, the non-academic society and uniformed brigades. Each subclass will define the main class and make it easier to do cross references. The class hierarchy applied for this development denotes an “is-a” association between one each other. The reason for this is because there are some classes which are actually the subclasses of another class in the hierarchy. This means, the properties of the classes will be inherited by the subclasses. Table 1: The EMS Ontology Class Hierarchy The EMS ontology has only a single inheritance as depicted in Table 1. The subclasses of subjects, syllabi and evaluation are common to all curricula concepts. Each slot value in the EMS design uses the primitive type (string, integer, etc). This approach is simpler than using instances or classes as the slots. Each slot has a single cardinality from which it represents every instance of a class and will only have a single value. Table 2 simplifies the terms used to describe the subject class. Domain: Education