Is it possible to find navigation elements that result in the best visitor experience? The answer is likely yes, and this question can be reformulated as one of the classic machine learning tasks.
Let's talk what are other classic machine learning tasks, and if and how they can be applied to content management
When your site contains more content items than can be displayed on a single browser screen, you need to prioritize some of them.
Let's say, you need to decide what items, and in what order, should appear in the main site menu.
How do we do that? There are several approaches:
- Go with the gut feeling (client opinion included).
- Take priorities from existing site (if we're working on redesign, upgrade, or migrate project).
- Analyze web server logs or/and Google analytics (or any other statistics analysis tool) to measure content performance, and reorganize content accordingly.
- Use more advanced technics like A/B (split) testing
What's common in all approaches, is that there is a gap between the change in menu, and analysis result. Yes, there are split testing tools that allow real-time optimization of page content, but a) they are difficult to setup and maintain, and b) they mean that the site content is managed by external party.
Luckily, the task of finding best performing menu structure can be formulated in terms of machine learning, and has well known solution.
I also would also briefly talk about other basic machine learning problems and their solutions, to see if they can be applied to the content management domain.
This session will not be Drupal-specific, but I will use Drupal terminology and sample site to explain machine learning applications.
