Meeting Minutes 08/01/2003

Machine learning project

1. Xiaoming has increase the number of web pages in each WAI category from 30 to 100.
2. He has used C5.0 decision tree induction algorithm to learn the decision rule from data
3. The ROC curve for C5.0 decision tree indicate it is a good candidate to differentiate double A and triple A conformance
4. There are two possible ways to combine the resutls of single WAI score and decision tree. one is use only the variables selected by the decision tree. The other is to use the information gain for each variable calculated by the decision tree algorithm.

Web complexty and Web accessibility

1. Al has extracted tags from the health consumer information websites
2. Correlation between the uniqueness of tags and total number of tags and the web accessibility barrier score are examined.
3. Both corrleation are signigicant
4. Next step will be checking subset the web page tages and tags relating to the web complexity as indicated by other papers.

Longitudinal study of web accessibility

1. Stephanie has query about 10 different websites using first archived web pages from each year since 1997.
2. She use the alexa to get the websites. excluce the one does not have pages in each year
3. Need educaiton websites and government websites. Will get it from AAU and topgovermentsites.com.
4. Will do some statistical analysis for the data. Starting from the plot to see any pattern.

Gateway server

1. Sajrif presented the architecture design of the gateway server
2. Bambang suggest to implement a simple one as soon as possible
3. Add transformation componenets into the architecture.