Random Ideas

Recommendation system

I'm thinking about implementing some sort of recommendation system for MoinMoin. This system should be able to present the user similar pages to the one viewed by him. I see two ways of implementing this recommendation system: a content based implementation or/and a collaborative approach

The content-based approach

In this case of text classification, the system searches for pages similar to those the user prefers based on a comparison of content

Problems

Content based systems tend to learn in a way that they recommend pages similar the the already seen ones

The collaborative approach

While the content based approach to text classification can be successfully applied to a single user, the collaborative approach assumes that there is a set of users using the system. In the collaborative approach the advice given to the user is based on the reaction of the other users. The system searches for users with similar interests and recommends pages this users liked. Instead of computing the similarity between pages, the system computes the similarity between users. The simmilarity of users is based on the comparison of the ratings they assigned to the same pages.

Generic MoinMoin Ideeas

Syntax

List syntax

One of the things I currently don't like in the MoinMoin synatax is that i am required to add whitespace to list markers

One example of the moinmoin syntax is:

 * Item 1
  * Item 1.1
  * Item 1.2
 * Item 2

The creole syntax would be much better