Idea: Optimise the evaluation of big rule sets

Here is project idea related to rules (for drupal 7) an interested student could pick up.

Currently rules follows a naive approach for evaluation rules, thus each rule and its conditions are check sequentially as specified. However when rule bases get big it would make much sense to optimize the evaluation of the rules. As something like the RETE algorithm doesn't apply to the module, I think the best thing to do would be reordering the rules in a decision tree, thus each condition is only checked once. A rules are already prepared and cached for evaluation, we could plug into this preprocessing step without hassle.

Tasks that such a project would probably have to deal with:
* Compute dependencies between conditions depending on their configuration. Thus most important detect equally configured conditions and their negations. Ideally it should also determine some further dependencies, e.g. if a node type is determined to be a page, it can't be an article.
* Implement the actual preprocessing of RuleSets based upon the determined dependencies and plug it in using a module.
* Find the best way to deal with weights in RuleSets. For the optimiser we need the rules to have the same weight, so they can be reordered. However currently Rules in d7 uses tabletrag.js for rules in sets, thus by default they would have different weights. (Still the rules reacting on events usually have the same weight, thus there the optimiser can do its work anyway!)

Of course I'd happy to mentor such a project.. :)