diff --git a/rfcs/text/0042-risk-score-extensions.md b/rfcs/text/0042-risk-score-extensions.md new file mode 100644 index 000000000..329dd3804 --- /dev/null +++ b/rfcs/text/0042-risk-score-extensions.md @@ -0,0 +1,124 @@ +# 0042: Risk field extensions + + +- Stage: **0 (strawperson)** +- Date: **2023-07-13** + + + + +This RFC seeks to extend the [existing risk fields](https://www.elastic.co/guide/en/ecs/current/ecs-risk.html) [(RFC 0031)](https://github.com/elastic/ecs/pull/2048) to support new/extended Risk Score investigation workflows. The workflows that this RFC intends to enable include all those described in 0031, along with the following: + +1. Risk Score Explainability + * We want to provide more insight into the anatomy of a risk score. The first (and simplest) way we intend to do this is by showing the documents (referred to commonly as Risk Inputs) that contributed to a particular risk score. Given that there may be a large number of these documents, we expect to have to choose a representative subset of these documents to persist along with the score (most obviously: top N riskiest inputs). + * Since we cannot realistically persist the _entire_ contributing document along with the risk score (let alone several), we intend to persist just enough information to allow one to uniquely identify those documents at a later point in time (i.e. during investigation/analysis of a risk score), along with any information that would not be present on the original document (e.g. the document's calculated risk score). +2. Categorical Risk Scores + * While the initial iteration of risk scoring ingested Detection Engine Alerts, we intend to expand risk scoring to include more data sources from multiple new categories of data. While we will still present a single risk score for most investigative purposes (composed of all these evaluated data sources), we believe that it will be useful to present individual risk scores _per category_ of data. + * These categories (and their definitions) are still being discussed [in this internal ticket](https://github.com/elastic/security-team/issues/5485), we currently know that categories will have the following traits: + * There will be a finite (<10) number of categories + * These categories' definitions may be _extended_ in the future to include new data sources + * Due to the above category traits, we need to come up with a naming convention for these categorical score fields that allows them to be extended without invalidating the existing field names. + + + + + +## Fields + + + + + +## Usage + + + +## Source data + + + + + + + +## Scope of impact + + + +## Concerns + + + + + + + +## People + +The following are the people that consulted on the contents of this RFC. + +* @rylnd | author +* @SourinPaul | SME / EA product manager + + + + +## References + +* [existing risk fields](https://www.elastic.co/guide/en/ecs/current/ecs-risk.html) +* [previous risk fields RFC (stage 3)](https://github.com/elastic/ecs/pull/2048) +* [internal risk categories epic](https://github.com/elastic/security-team/issues/5485) + +### RFC Pull Requests + + + +* Stage 0: https://github.com/elastic/ecs/pull/2232 + +