Observation Semantics 4 capabilities
Lesson 2 of 4 domains — 4 capabilities: how the information was observed.
This domain answers: How was this information observed — where, what scope, what method, what time span?
Source for this lesson
semantic-capabilities-reference.md, section "Observation Semantics (4 capabilities)"
1. The question this domain answers
Once you know what a value is about (Lesson 1), this domain asks how it was actually obtained: where does it come from, how much of the system does it cover, what technique produced it, and over what span of time. Two readings of "soil moisture: 22%" can mean very different things depending on the answers to those four questions — one sensor vs. a district-wide average, sensed vs. modeled, right now vs. a rolling weekly average. This is the domain with the most internal tie-breakers, because these four capabilities interact with each other constantly.
2. The four capabilities
| Capability | Question it answers | Reference doc's examples |
|---|---|---|
| Observation Point | Where does the observation's referent physically or logically sit? | Luminaire; cabinet; line head; street surface; pump; valve; pipeline; root zone; weather station |
| Observation Scope | What portion of the system does this one value cover — one asset, or many combined? | Single asset; group of assets; cabinet; street; irrigation zone; district; entire municipality |
| Observation Method | What technique produced this value — sensed, or computed/derived? | Measured; estimated; inferred; predicted; aggregated; simulated; model-derived; proxy measurement |
| Temporal Semantics | What time span does this value represent? | Instantaneous; observation interval; sampling period; aggregation window; historical; forecast; prediction horizon |
3. The tie-breaker: two ways to "aggregate"
"Aggregated" is the single most overloaded word in this domain, because you can aggregate across space or across time, and they land on different capabilities:
- Aggregating across assets/locations (e.g. one number combining 40 luminaires into a district average) is primarily an Observation Scope question — what portion of the system does the number cover — with Observation Method as a companion, since the aggregation technique itself is still worth recording.
- Aggregating across time (e.g. a rolling 24-hour average from one sensor) is a Temporal Semantics question, even though it was also computed rather than directly sensed.
Rule of thumb: ask what got combined. Multiple places → Observation Scope. Multiple readings over time → Temporal Semantics. Observation Method is never Primary for a pure aggregation-over-time question — it only rides along as a companion when the computation itself matters.
4. Fresh example — a district lighting average
| Sub-observation | Primary Capability | Companion Capability | Why |
|---|---|---|---|
| Covers all 40 luminaires in District 5 | Observation Scope | Observation Method (aggregation technique) | The number is a space-combination across assets — the defining fact is how much of the system it covers. |
| Over the last hour | Temporal Semantics | — | A separate question: what time window does the value represent, independent of how many assets fed into it. |
Notice this one observation splits into sub-observations along two different axes at once (space and time) rather than one obvious split. That's normal for this domain — always check both axes before deciding you're done.
5. Observation Point — a preview of Lesson 3's tie-breaker
The reference doc's guidance for Observation Point: "when the same device name could also answer 'who reported it,' default here as primary with Provenance as a companion, unless the sub-observation specifically contrasts sources." In other words, a device name usually tells you where something is (Observation Point) more than it tells you who supplied it (Provenance) — Lesson 3 covers exactly when that flips.