In [[automated submission web directories]] human editing or [[moderation]] is replaced with the automated processing which requires special techniques for the website content categorization and filtering. This task is similar to the [[document classification]] which is one of applications of [[machine learning]] techniques.
{{inappropriate tone|section|date=April 2009}}
Creating automated submission web directories helps to avoid long delays in approving submissions. Yet this is a challenging task because the human editing process is replaced with an automated decision process, which should be reliable. Basically there are two major aspects in automation editing which should be addressed:
#assign a right category to a website
#filter out undesirable websites
The first one requires a mechanism which classifies the [[web content]] and decides which category (or most relevant categories) among a given set of categories this website belongs to. Two major features which constitute websites are the text and [[images]]. Although a substantial progress in [[image processing]] has been achieved in the last decades it is still far from the level when complex images can be successfully, reliably and promptly classified. Text or [[document classification]] is much more advanced. Since usually a set of categories in web directories is predefined [[supervised document classification]] may be applied here.
The filtering (or imposing restrictions on submitted sites) aspect is also very important. For example, the same link is not allowed to be listed twice. Along with other simple tasks this one is solved using straightforward algorithms. Much more complicated task is to filter out sites with specific content, such as inappropriate content including legal and illegal pornography, racial hatred, violence, etc.
Using Machine Learning methods is the way to tackle these problems. An important feature is a reliable category recommender system which uses [[machine learning]] methods for content classification in both category suggestion and filtering.
Of course, automation is never perfect. However one very important for SEO purposes advantage of using automated submission based on machine learning methods over human editing is better category suggestion for the website content. Because of [[statistical]] nature of machine learning, automated categorization may result in more relevant category suggestion in terms of text interpretation by [[search engines]] than what the human perception may suggest.
