Where to ask about Bayes classifiers for spamming detection? How to draw a line between CS and AI?

Question

I wanted to ask which method for spamming detection would be more suitable for certain scenario, Naïve Bayes or Artificial Neural Networks, but then I've found 4 sites where Bayes classifiers can be on-topic:

• Stats.SE -> 316 questions (1/3 unanswered),
• DataScience.SE -> 17 questions (4 unanswered)
• Math.SE -> 35 questions (half unanswered),
• CS.SE -> 13 questions about bayesian statistics (no specific tag for classifiers),
• CSTheory.SE -> 17 results on Bayes word, no tags for it at all.

But then I've found that Wikipedia page for Naive Bayes classifier says:

In machine learning, naive Bayes classifiers are ...

It doesn't say in statistics as oppose to 'sampling' (as example), where we can read:

In statistics, quality assurance, and survey methodology, sampling is

It says specifically in machine learning, not statistics, which further more, the machine learning is a sub-field of Artificial Intelligence (as suggested in this paper or here):

Machine learning, a branch of artificial intelligence, is about the construction and study of systems that can learn from data.

To make it more tricky, the machine learning is also a subfield of computer science (as per wiki page). But on the other hand I'm not interested discussing math equations which I'm seeing a lot on CS.SE, because I'm looking for more practical oriented answers, not theories. Secondly they've not specific tags for these classifiers. I'm also not studying this topic (on academia level), but I'm doing this as a hobby, or more specifically, investigating practical problem solutions in my app stack.

Based on above logic, does it mean asking AI.SE is the most suitable place to ask practical questions about spamming detection using Bayes classifiers?

Answer 1

I think most questions about Naive Bayes classifier belong on stats.SE or data.SE.

It is part of data mining and data science, and it is probably on-topic on data.SE. Some examples of posts on data.SE that appear to be similiar to the question you want to ask:

• How can I classify text considering word order, instead of just using a bag-of-words approach?
• Spam detection in social media. This question uses another method, but Bayes is suggested in an answer.

A question on stats.SE that is similiar: Simple text classifier: classification taking forever?

The point is, a theoretical question about the Naive Bayes classifier will probably belong on stats.SE since it involves probability and statistics. An applied question would probably be better on data.SE.

Also, your wiki argument is not a really good one, since anyone can add or remove such sentence. Here is one from an other language that start with such sentence:

En teoría de la probabilidad y minería de datos, un clasificador Bayesiano [...]
(In the theory of probability and data mining, a Bayes classifier [...])