|   DESCRIPTION    |    SOM    |    DEMONSTRATION    |

This project is to build a tutoring application for Self-Organizing Maps as a part of Interactive System Design course. The goal of the project is to develop a Web-based system that serves real users. This project should bring knowledge of interfaces, interface design and evaluation, and programming Web-based interactive systems together. The system will focus on educational aspect along with interactive and visualized features.

Why a Tutorial Application for Self-Organizing Maps?
From a data mining course with Dr. Hirtle, we had an experience of having seen students having difficulties in understanding the concept of Self-Organizing Maps as a part of clustering concepts. Most of confusions were from understanding concepts of clustering and visualizing it into maps. Therefore it is significant to give a point to the difference between map and graph. In addition, the training steps are a part that makes students confused which requires representation of steps changes and algorithm processes.

System Design Focus
The pages of the project has identical design so that users can get used to the system easily. Also the design is kept simple for the same purpose. With the results from the interview with Dr. Hirtle, the interactive aspects are very important for the tutorial system because students can play with it and get better idea of what SOM is. Considering interactive aspects of the system, some features were suggested from the interview such as a control over iteration to let users comprehend the process of SOM algorithm, a funtion of comparing multiple seeds.
It is also suggested that in terms of building a tutorial application, it would be very helpful to have multiple representation features for the same data because students can see how SOM is working and can understand by comparing different representations. This feature will be very useful since the location in the result of SOM in the map does not have any meaning of the location, and since the result of SOM just represents just the similarity of data.
In addition, it is also pointed that student usually get lost in interpreting the result of the SOM map. Therefore including some definition of cluster or explanation of the result of the SOM map to the tutorial tool will be useful for education purpose.

As summary, the following list is suggested features or functions for the tutorial system in design point of view.

• Showing data information for individual cells such as weight vectors
• Multiple representations of same data set : mirrored images of the map
• Control of iteration and random seeds
• Cluster definitions and map interpretation
• Explanations on the each step of the process