Classifying ASNs in our ASN database
Let’s start with an example right away.
AS Organization name: Signet B.V.
AS Organization domain: signet.nl
When you look up the information of this AS organization, which is associated with multiple ASNs, you expect to get one single AS-type declaration, however, when you take a closer look at the ASN database or our IP to Company data.
You can see that the Signet B.V. organization is associated with three different ASN types:
- Hosting
- ISP
- Business
| ASN | DOMAIN | NAME | TYPE |
|---|---|---|---|
| AS31586 | http://signet.nl/ | Signet B.V. | hosting |
| AS49685 | http://signet.nl/ | Signet B.V. | isp |
| AS39700 | http://signet.nl/ | Signet B.V. | business |
Should an AS organization be associated with one AS type?
This AS organization, like many, provides multiple distinct services across its multiple ASNs, so we are classifying them into multiple categories.
Signet B.V.’s website shows they provide hosting and ISP services and operate their own business entity service. For that reason, it is expected to categorize the different ASNs of one organization into multiple categories.
How do we categorize ASNs?
Even though ASN data is largely collected from public data sources like WHOIS data, we infer the AS type data. We use a statistical model that compares the behavior of specific ranges and compares them to a known baseline.
Internally, the AS prediction database model looks like this:
| id | type | business | education | hosting | isp |
|---|---|---|---|---|---|
| AS141039 | isp | 0.019 | 0.0 | 0.127 | 0.854 |
| AS147049 | isp | 0.027 | 0.0 | 0.035 | 0.938 |
The AS type selection scores imply that the category ‘type’ is not necessarily mutually exclusive, but rather the API selects the “best”. If AS-type prediction is considered a critical factor in your business decision, you can request the ASN prediction database or our AS/Range Scores fields as part of our enterprise custom data solution.
IPinfo services referenced:
API
Database