Now we are focused on creation the intelligence system of courses management. We aim to generate relationships between courses automatically. Every course from the directory can be mapped to other courses as an alternative or a prerequisite or s similar subject course etc.
Such type of relationships between courses will be used to generate personal education path automatically and individually for every online student.
For now we release one piece of this work. It is “See also” listing for a course.
On a course description page we will display the list of most relevant courses related to the courses. But not exact alternatives as we have different listing for a course alternatives.
To see how it works just open courses in the directory. Now this function will work for more then half of courses in the directory.
Recently we added support of mapping courses in our directory to categories. Now it is possible to view categories list on the site Categories of Online Courses.
It is difficult task to map 5400 courses to categories. It is not possible to do this manually. We asked your users to help with mapping courses to categories . And we are grateful for your help.
How we do categorization of MOOC courses.
On first step we mapped categories for part of courses manually (and our visitors helped with this).
Then we used some categorization techniques to map rest courses automatically. We get descriptions of already mapped courses and used them to train N-Gram-Based Text Categorization tool. For every unmapped course we used trained tool to find best category. For now this works successfully in 80% cases. Soon we expect to get better results and then new MOOC courses will be classified automatically.
Why categorization is not easy to do?
MOOC vendors don’t provide some standard description for their courses. Not all vendors have courses categorized. Even if a vendor has categories and the list of categories is different from vendor to vendor. We aimed to create unified list of categories.