Set up automatic processing of record linkage suggestions #221

New issue

Closed

opened 2024-09-27 01:02:47 +00:00 by fricklerhandwerk · 1 comment

fricklerhandwerk commented

2024-09-27 01:02:47 +00:00

(Migrated from github.com)

For security team members to be able to triage effectively, we need pre-computed suggestions for record linkage:

As new CVEs are ingested, match them with existing derivations.
As new derivations are ingested, match them with untriaged CVEs.

Open questions:

Should newly ingested items also be proposed for getting added to existing linked records?

For security team members to be able to triage effectively, we need pre-computed suggestions for record linkage: - As new CVEs are ingested, match them with existing derivations. - As new derivations are ingested, match them with untriaged CVEs. Open questions: - Should newly ingested items also be proposed for getting added to existing linked records?

RaitoBezarius commented

2024-10-03 14:14:00 +00:00

(Migrated from github.com)

It's pretty rare to introduce a new derivation for which we already have a CVE in (except for historical open CVEs, which can be dealt with a one-time matching).

Matching a new derivation with untriaged CVE is touchy because our CVE database can go back to 1999. This deserves a bit more thought to do it properly and not dealing with it doesn't reduce the problem because the way we care the most is: we already have the derivation ingested and we have a new CVE. It's pretty rare to introduce a _new_ derivation for which we already have a CVE in (except for historical open CVEs, which can be dealt with a one-time matching).

👍 1