Skip to content

learnweb/moodle-mod_ratingallocate

Repository files navigation

moodle-mod_ratingallocate

Build Status codecov

Module which lets you add an activity to courses, in which users can rate choices. You may then distribute the users fairly to the choices by maximising overall 'hapiness' in terms of ratings. This may be an alternative to the choice activity or other first-come-first-served plugins.

This plugin is based on previous work by Stefan Koegel and Alexander Bias, University of Ulm.

Installation

This is an activity plugin and should go into mod/ratingallocate. Obtain this plugin from https://moodle.org/plugins/view/mod_ratingallocate.

Usage

Add an activity instance. Set mandatory parameters. These are timespan, in which users can give ratings, and the strategy, which form will be presented to the users to rate. Next you can add choices, which the users will have to rate later on. After the rating period has finished, you can allocate the users automatically or manually. Upon publishing the results, users will be able to see which choice they have been allocated to. For more information please visit the moodle wiki.

Moodle version

The plugin is continously tested with all moodle versions, which are security supported by the moodle headquarter. Therefore, Travis uses the most current release to build a test instance and run the behat and unit tests on them. In addition to all stable branches the version is also tested against the master branch to support early adopters.

Algorithm

This module uses a modified Edmonds-karp algorithm to solve the minimum-cost flow problem. Augmenting paths are found using Bellman-Ford, but the user ratings are multiplied with -1 first.

Worst-Case complexity is O(m^2n^2) with m,n being number of edges (#users+#choices+#ratings_users_gave) and nodes (2+#users+#choices) in the graph. Distributing 500 users to 21 choices takes around 11sec.

Changelog

The changes for every release are listed here: https://github.com/learnweb/moodle-mod_ratingallocate/wiki/Changelog.