How to determine the affiliation of domains?

Question

Domain Name System

We have a million domains related to legal entities. It often happens that a dozen legal entities are affiliated at the level of it-infrastructure, have a common mail server, or something else.
We need a list of parameters that can be obtained, knowing the domain.
It occurs to me 1) Determine ns, mx and other domain records through dns queries. 2) Determine the mail server.

Answer 1

Yandex sharpened the affiliate filter by a bunch of parameters, but you only want to find affiliates by domain? )) why then Yandex could not determine by one domain ... I studied the content on sites, cms, contacts, etc.
But at the same time, with a margin of error, IMHO, you can try to find a selection of sites that most likely belong to the same company.

The main factors by which the Yandex search engine can determine that one site is an affiliate of another may be the following (in percent the degree of risk is indicated):
Registrar match -16% If the
domain registration dates are very close -22
%
sites have the same CMS -26%
Sites are located on the same hosting -32%
A large percentage of matching donors in the reference mass -49%
Sites are in the same subnet (three octets in the IP address are enough) -59%
Matching information about the organization on the site and in the reference -68%
The presence of relinking between sites, that is, if the sites link to each other -72%
Matching site owner data -79%
The same content (meaning not the uniqueness of the texts, but the one direction of the site) -93%
The presence of the same contact details (email, skype, etc.) -99%
The coincidence of the address, phone number and name of the organization -100%

Source - https://seonomad.net/article/affiliat-filtr-yandeksa
, that is, in your case it makes sense to look at the items marked in bold.
if many parameters matched (according to the degree of risk), then with a high degree of probability it can be assumed that the sites belong to the same office.
Moreover, it is not a fact that they will be affiliates in the full sense of the word.
It's just that the office can have 15 sites on different topics, regions, etc. - a kind of diversified business. That is, if you can find out by domain, then certain sites that most likely belong to one office, and the subject of sites, their keys, etc. You can't find out from domain records.

Answer 2

None of the above can be attributed to direct evidence. Legal entities can be hosted in the same data center, use mail hosting services from each other - but this does not make them affiliated. It is clear that if the DNS records of one company lead to a range of addresses belonging to another, it is obvious that there are some relationships between them, but to call it affiliation right off the bat, IMHO, is too bold.
And given that many people use services like CF, Google mail, etc., the task becomes even more difficult.