In our research activity this week, Jessie Chuang offered an idea, promoting semantic interoperability by requiring consistent identifiers for vocabulary.
[The] xAPI spec has only guaranteed data structure interoperability. To guarantee semantic interoperability for xAPI data sent by different systems, there should not be multiple ids for one semantic meaning. There are a lot of considerations in vocab… all documented in [ADL] publications.
Additionally, Jessie includes links to
- AcrossX vocabulary
- ADL’s Vocabulary Specification
- ADL’s Vocabulary Primer
- Guidelines for IRI Design and Persistence –
- Controlled Vocabulary Considerations for the Experience API (xAPI) –
Consistent identifiers support semantic interoperability.
First, I believe that practitioners share a goal when we create data and use xAPI to do so. That goal is, I believe, to create data that’s a) immediately useful in terms of giving us insight and b) usable by different systems in different contexts. That’s what semantic interoperability means, in laymen’s terms.
Therefore, if we agree that’s a good goal, then we can dig into the specifics of the suggested requirement. For my part, I think this is a pretty good idea, though a number of questions cross my mind in considering it as a conformance requirement:
- How do we reinforce that requirement so people creating data can easily do the right things?
- How do we test for that requirement to reinforce it?
- What would we test to meet that requirement?
In addition, I can think of other, more pragmatic questions.
- How much data that’s currently mulling around in LRSs around the world would not conform to that requirement?
- How many current systems, be they LRSs, Learning Record Providers or “other” things currently implemented and/or on the market could conform to that requirement?
- What would we suggest doing about it?
It’s that last question that ultimately we have to tackle. That’s why it’s important to distinguish the nuance of requiring consistent identifiers from the goal of semantic interoperability. There might be other ways to support that goal.
Ultimately, I can’t think of any better approach than consistent identifiers off-hand, but I remain open to learn something new and to take suggestions.
What do you think? Comment below and please help break down Jessie’s idea into more granular requirements.