Can a discrete event include processes?

Yes. A delivery can contain driving, waiting, and handling activities.

Does subdividing time create more completed events?

No. Event identity and completion conditions determine the count.

Is this the same as discrete-event simulation?

The ideas are related through event modeling, but simulation often represents scheduled changes at particular times; the supplied term concerns structured occurrences and process descriptions.

AI Glossary / Graphs and symbolic AI / Advanced

Discrete events

Definition

In knowledge representation, discrete events are occurrences with a definite structure and identifiable boundaries. Their complete description is not generally preserved when an arbitrary smaller portion is considered. This entry follows the supplied AIMA meaning, contrasting bounded events with ongoing processes, rather than treating every event as an instantaneous computer notification.

What it means

Consider completing a delivery from a depot to a customer. It includes an origin, journey, and arrival. A five-minute portion halfway through that journey is part of the delivery, but it is not itself another completed depot-to-customer delivery.
By contrast, under a suitable idealized description, driving can hold throughout an interval and its smaller subintervals. AIMA calls categories with this subdivision property process or liquid-event categories. The distinction concerns how an occurrence is described, not whether physical time moves in discrete jumps.
A bounded event can contain smaller events and ongoing activities.

How it works

A representation identifies the event, its participants, its type, and its temporal extent. An application might record a delivery identifier, vehicle, package, origin, destination, start time, and completion evidence. Separate assertions describe what occurred during the event and which conditions held before or afterward.
Temporal relations support questions such as whether two intervals overlap, whether one precedes another, or whether an interval lies within another. OWL-Time supplies vocabulary for instants, intervals, durations, and their relations. That temporal vocabulary does not itself decide whether a delivery was successfully completed; the application must define its event semantics.
When splitting an interval, preserve the distinction between a whole event and a portion of its activity. Whether a description holds on every smaller interval depends on that description and the chosen granularity. An interruption can create a failed or interrupted attempt without satisfying the original completed-event category.

Three practical examples

Delivery records: A logistics knowledge base distinguishes a completed delivery from a vehicle being in transit. Location updates can describe movement without proving that the package reached its required recipient.
Document workflows: A system distinguishes an approval procedure from reviewing a document. Several periods of review can occur without the final approval decision. An AI summary should not infer approval merely because review activity was recorded.
Manufacturing records: A production system represents a batch run as a bounded occurrence with specified start and end conditions. Heating or mixing can occur within it. Observing one of those activities does not establish that the whole batch run reached its required outcome.

Think of it this way

Think of a complete musical performance and a short passage within it. The passage belongs to the performance, but listening to that passage is not listening to the whole performance. At the same time, the description “music is playing” may apply throughout both intervals at the level of detail being modeled.

Worked example

Suppose delivery D17 starts at a depot at 09:00 and completes at a customer's receiving point at 09:40. For this example, define a completed delivery as one identified transport from that depot to that destination, ending with a recorded handover. Assume the handover is verified at 09:40.
The complete event lasts 40 minutes. Divide its interval into four ten-minute portions: 09:00-09:10, 09:10-09:20, 09:20-09:30, and 09:30-09:40. Each portion is part of D17. None is a separate completed depot-to-customer delivery under the stated definition.
A query asking how many completed deliveries occurred therefore returns 1, not 4. A query asking how many ten-minute portions partition the recorded interval returns 4. The answers count different things.
Now suppose uninterrupted driving is recorded from 09:05 to 09:25. It lasts 20 minutes. Within the model's granularity, “driving” also holds from 09:10 to 09:15. This five-minute subinterval preserves the process description even though it does not preserve the complete-delivery description.
Finally, change the scenario: the vehicle stops at 09:20 and the package never reaches the customer. The record supports an interrupted transport attempt, not a completed D17 delivery under the defined rule. Twenty minutes of observed activity is not half of a completed delivery for counting purposes. A separate progress measure may report partial work, but it must not silently replace the completion criterion.

Illustrative case study

Fictional Willow Dispatch builds an AI report from vehicle updates. Its first summary counts every movement interval as a delivery event, so a journey divided into several telemetry segments appears as multiple completed deliveries.
The team introduces a persistent delivery identifier and distinguishes interval observations, transport attempts, and verified handovers. Reviewers test completed, interrupted, and still-open journeys. The reporting logic counts completed events using the handover rule while retaining movement segments for duration and route analysis.
They can describe ongoing activity without asserting completion and can count a single bounded event without losing its internal history.

Common misconceptions and mistakes

“Discrete means zero duration.” A bounded event may last minutes, hours, or longer.
“Every subinterval is another event of the same type.” A portion of a delivery is not automatically a completed delivery.
“An activity log proves the intended outcome happened.” Completion requires the evidence specified by the model.
“Liquid event means a physical liquid is involved.” It names a subdivision property of the process description.
“An interrupted occurrence cannot be represented.” It can be represented under a suitable attempt or interruption category.

Business use

Define the unit being counted before building metrics or AI summaries. State the start condition, end condition, event identity, and required completion evidence. Otherwise, intervals, messages, attempts, and successful outcomes may be mixed into the same number.
Preserve partial observations and uncertainty. A missing handover record may mean completion is unverified rather than disproved. Distinguish unknown status from a known failed attempt, and retain the evidence that supports later corrections to an event's description.

Check your understanding

Why is the completed-delivery count 1 rather than 4?

Answer: Four time portions belong to one identified completed event.

Why can “driving” describe a smaller interval?

Answer: Under the stated continuous-activity assumption, that process description remains applicable.

What does an absent completion record establish?

Answer: It establishes missing evidence, not necessarily that completion never occurred.

Related terms

Event
Knowledge representation
Ontology
process
temporal interval
event identity
temporal reasoning
completion condition

Learn next

Choose one workplace event and define its boundaries and completion evidence. Separate its constituent activities from its outcome, then test how the representation handles interruption and missing observations.

Event

Sources

AIMA glossary, discrete events and process-category meaning: https://github.com/aimacode/aima-glossary/blob/master/glossary.md Stony Brook University, AIMA knowledge-representation teaching slides: https://www3.cs.stonybrook.edu/~sael/teaching/cse537/Slides/chapter12_KR.pdf W3C, OWL-Time temporal vocabulary: https://www.w3.org/TR/owl-time/

Previous: Discrete distribution Next: Discrete feature