Integrating Agents into Teleteaching-Webportals

The aim of the project is to combine an existing agent system for mobile agents with a web portal in practical use for teleteachingelelearning. This combination of web portals and agents yields a number of advantages for users, such as automatic notification of different events, autonomous handling of simple tasks and supporting the teaching/learning process. The main issues in the project are the integration of agents into web portals for teleteaching, implementing agent- based support for coaching processes, identifying security issues particular to this combination, and evaluating this use with regard to acceptance and utility for learners and teachers. In addition to this, legal aspects of teleteaching web portals in general and those arising from combining them with agents in particular (data protection, liability for acts of agents, copyright, links) will be investigated in this project.

The aim of the project has been to integrate an agent platform with an E-Learning platform. Both platforms already existed at the institute for information processing and microprocessor technology and have been combined in the context of the project. Additionally some transfer of agent technology directly into the learning platform has taken place. Communication between both systems uses the open SOAP standard, which enables other programs to interface with either system as well, regarding both data and functionality. The most important aspects and results of the project are discussed below. Because of the specific and constitutive property of multi agent systems, namely that agents are independent from each other, it is necessary for the learning platform to address the "correct" agent for each request. To avoid having to create a complete image of the agent society within the platform (danger of inconsistencies), all communication is routed over a dedicated "gateway" agent. This might lead to performance problems, so redirection for SOAP requests was invented and implemented. In this way the advantages of a single contact point are retained without the danger of a bottleneck. Because of the temporal independence of agents from user actions, these are especially suited for asynchronous tasks. This includes e.g. identification of user interests based on their actions within the learning platform. For this it is partly necessary to inspect large amounts of data and therefore this process can be scheduled to take place at times that the system is idle (e.g., during night). If, for instance, learning objects are not annotated with metadata, this must be derived by the agents automatically from the content. Based on this information they can then identify the areas of interest for users. This derived information is then used for instance for additional filtering of notifications on new messages posted within the platform (e.g. in forums). Another application of agents is support for awareness. In a graphical representation of the course, additional information is introduced so learners can asses their own progress in comparison to other learners. This graphical representation shows which parts have already been visited as well as where other (anonymized) learners currently are. This area is being enlarged at the moment to allow, for instance, learners to contact others in their near vicinity. Additional integration of agents took place in the area of notifications on events, especially new postings in discussion forums. For this a rules engine is used, which receives the events as well as information (e.g. the interests) from the users` personal agent and then triggers configurable responses. These notifications are shown on the personal page within the learning platform, but can also be transmitted by E-Mail or SMS. These additional functionalities have led to an increase in the popularity of the learning platform, with additional institutions currently evaluating it for adoption in their own distance learning courses. This results additionally in the creation of more learning material and reduced the need for introducing students to the platform.