Internationalization in STIX 2.1

Hey all,

One of the topics we’ve been grappling with for awhile in STIX 2.1 is internationalization. We want STIX content, specifically, to be translatable into different languages so that you can have publishers creating content in several languages, third party translation services for STIX content, and identification of which language specific content is in.

We’ve made a lot of progress towards that end. Here’s what we have:

Each top-level object will have a field called `lang` to indicate what language that content is in. That means that a campaign object could be published with a `lang` field of Japanese to indicate that the name, description, and other human text fields are in Japanese. (https://docs.google.com/document/d/1HRVFn2kAxBOTMbEb3KRu8tjMoHm-KRAI-2R8CTzGil4/edit#heading=h.xzbicbtscatx scroll down to `lang`)
A “granular language markings” capability, so that you can indicate that certain fields in the object are in English, while others are in Japanese. (https://docs.google.com/document/d/1HRVFn2kAxBOTMbEb3KRu8tjMoHm-KRAI-2R8CTzGil4/edit#heading=h.l6edgya0tyjq see new `lang` property on granular markings)
A separate “language-content” object that allows you to provide additional languages (translations) for that content. So you could publish a campaign in Japanese and then provide a language-content object with that same text in English. (https://docs.google.com/document/d/1HRVFn2kAxBOTMbEb3KRu8tjMoHm-KRAI-2R8CTzGil4/edit#heading=h.cfz5hcantmvx)

One thing we have no decided yet is whether each top-level object in STIX MUST have the `lang` field populated on every object or SHOULD have the `lang` field populated on every object (not including bundle, observed-data, and sighting, which don’t have translatable fields). There have been previous discussions on Slack, e-mail, and at the face-to-face…you should be able to dig through the mailing lists to find it, the most recent one can be found in the Feb-Mar 2017 archives:

It seems like at this point we just need to have a ballot to resolve this. So, I move that the TC open a ballot to determine whether we feel the `lang` property should be required on each top-level object or optional on each top-level object (i.e. STIX Domain Objects, STIX Relationship Objects, data-marking objects).

John

cti message