Intelligent Learning Environment for Course Telematics
This project aims at developing processes for transforming text/paper based masters level course material into an interactive distance-learning environment based on the Internet.
Project Objectives 
Project Methodology 
  Information Collection Design 
  Baseline Version 1 
  Hypertext Version 2 
  Multimedia Version 3 
  Information Agents Version 4 
The process of developing hypertext features, multimedia content and intelligent agents into a student-centered information system will be studied. The course materials produced will be used on a number of existing MSC courses. The practical experiences of students on those courses will be developed as case studies. Included in the project will be the identification of changes required to implement educational information systems of this type in different countries/ cultural settings.
As important as the identification of successful implementation strategies will be the study of failures and attempts to understand why apparently sensible approaches can fail. Each of the pilot studies using prototype course provisions will be assessed considering a wide range of attributes of the study process. 

Project Objectives    

  • Develop the RDBMS module into a basic Internet form (Version 1) 
  • Develop the RDBMS module into a hypertext document form (Version 2) 
  • Develop the RDBMS module into a multimedia enhanced form (Version 3) 
  • Develop the RDBMS module into a information agent enhanced form (Version 4) 
  • Observe and document the process and options of each stage of transformation. 
  • Evaluate the effectiveness of the process used and quality of the products produced 
  • Specify a process to guide new developers saving their time and avoiding problem areas. 
  • Disseminate the details of this process and experiences gained completing the project to a European wide audience of commercial and academic distance learning developers 
Methodology of the Project  

Information Collection Design   
The information to be collected to evaluate the effectiveness of each of the Internet Versions will be determined using a Goals/Questions/Measures (GQM) analysis at two levels. Each of the partners will apply the same strategy for the GQM analysis in their different countries. The intention of this is to bring the results together to determine a core set of metrics which will be applied across each of the pilot study locations and to gain insight into possible cultural differences between sites - thrown up by different social and educational systems operating between the partners of the consortium. As a result of this stage of consideration a common vector of project evaluation metrics will be agreed and applied to each of the pilot study evaluations throughout the use of each of the 4 versions of the Internet module being referenced by students. 

Baseline Version 1   
The initial analysis is in order to set the baselines upon which the rest of the evaluation will be based . The existing (paper-based) unit wil be studied and how students perform using it. The initial study of the unit that has already been run will be used as the basis to inform the design of the other evaluation studies. In addition to this detailed analysis into the nature structure and content of unit itself will be performed at. The outcome of this will be a conceptual map of the course. The analisis will consider structure and inter-relation, learning outcomes, learning pre-requisites , type of learning required. Rather than use a formal, linguistic model the analysis will aim to be at a cognitive level explaining what is being taught and what the student is expected to learn, how it is being taught, how the student is expected to learn the concept (new concept, generalisation of old concept, specialisation of old concept, analogical mapping) and a taxonomic classification of the properties of the concept being taught. The result will be a detail map of the course unit which informs on its structure and content. Understanding how people cope with the existing paper-based material is clearly also very important so as it provide the basis for comparisons with subsequent electronic versions. The baseline version will then be transferred to the WWW as a set of electronically scanned pages to form version 1. Minimal hypertext linking will be incorporated at this stage. 

Hypertext Version 2  
The development of the second version of the database module on the Internet will involve determining a systematic process for introducing hypertext links into the baseline version of the module. 
Internal module hypertext links 
External module hypertext links to sources beyond the database module 
There are considerable difficulties in understanding where the balance must be drawn between a link-rich environment and a "crowded" linking environment where important hyperlinks are hidden in the mass of links which have less importance. The INTELLECT process in this area will concentrate on both the practical aspects of link identification and in developing a classification scheme for link environment will result and it will be evaluated considering the results of the pilot studies conducted at each user site. 

Multimedia Version 3   
Introducing multimedia into a hypertext document requires many considerations relating to the: 

  • media type used to convey information 
  • design of WWW pages (Human Computer Interface) 
  • nature of dialogues and interactions which can take place 
  • style 
  • practical space/time efficiency of building a hypertext document/environment 
This workpackage will require alternative forms and interactions to be tried and developed for the same area of the database module to identify a suggested form for users to consider and to identify the areas of choice available. This is a difficult area of the project and can only be explored to a very limited extend - considering the range and wealth of material published relating to HCI design and form. However, the main simplification which help us here with this project is that it is specifically focused on the application of implementing a masters level module on the Internet. 

Information Agents Version 4   
Artificial Intelligent Agents will be developed as adjuncts to self-paced standalone distance material. A JAVA based interface engine will be developed, allowing knowledge based programming of the individual agents. The goals of this module will be to 

  • improve the interaction and more nearly approximate to human discourse 
  • provide virtual "peers" with which the student can interact and collaborate 
Ther use here of AI is in a different vein to intelligent tutoring systems and their seeming intractable problems. Instead the idea is to take a limited part of the interaction and look to animate it with a software agent. The agents are small characters that users encounter as part of their learning experience. The agents are different and take on different roles naturally as they occur in the experience.Initially the following roles will be considered: 
Guide: An information Provider that students can turn to when they want. It will have a knowledge of the pages content, and also of the overall structure of the unit, in order to fulfil a navigational role. 
Co-worker: One of the problems with distance education is the isolation it implies. the lack of peers with which to interact is a major problem. Here we look to develop Co-workers, so that students can undertake collaborative assessments and projects, while interacting with a fellow "virtual" student. 
Boffin: Someone who the student can turn to if they want further help. Alternatively, the student can ask the boffin to aks questions of them in order to test their own knowledge and understanding. 
Fellow Web Surfer: To provide support, company and empathy. They aim to add separate characters to cyberspace and thereby improve and expand the interaction. They will have their own special interests and will be able to talk knowledgeable about subjects unrelated to the course or the Web. 
Typically agents will be located within pages. Their knowledge and ability to behave will be defined strictly within the confines of the page in which they are located. This constrains the types of behaviour they would need to perform and makes it a tractable problem using current AI technology. Thus the intention is to use a simple generic interface engine with pattern matching, augmented with model of simple beliefs/desires/ intentions. There will be an issue of keeping the code size of these agents small as this will effect JAVA applet download times. If this is going to be a problem local optimisation will be performed to reduce size.