add example showing non-deterministic "sequence"

[greenTara]

Per request, I have added an explicit example (Location_TempC_Minute_MergedCities.json), corresponding to the usecase of a syntactic merger of three streams of temperature data at particular locations to create one stream.

The original streams are in the form where the name of the graph is an IRI, and the graph is not included explicitly in the stream.

The merged stream simply combines the stream elements of the original streams, without change.

The observation times coincide, and so a sequence ordering of the timestamp triples cannot be uniquely determined.

If merger required creating a new timestamped graph combining all observations at a given time, then this would require dereferencing the graph names, and I don't think this would be an acceptable alternative.

[greenTara]

I have added a new example for illustrating issues about bnode scope in the semantics of RDF streams. In the Location_Subject.json example, the stream contains observations of unknown subjects (bnodes _:Subject_A _:Subject_B) at certain locations (X, Y), with observation time used as the timestamp.

Regarding bnode scope, the question is whether the bnodes representing the unknown subjects should be consider to represent the same entity throughout the stream, or should they be considered (possibly) different subjects. If one takes the view that the intent in this stream is to track the movement of entities over time, then the semantics should allow sharing of bnodes between timestamped graphs in the stream. Without such bnode sharing, the subjects would need to be skolemized in order to represent such tracking information.

(Thanks to Pat Hayes for providing this example, in a private conversation.)

[greenTara]

Further, here are some examples with interval timestamps (Location_Schedule_Harris...json), intended to represent a case where there is not a total order of the temporal entities. In this case, the order is based on Allen interval relations, where if interval A precedes or meets interval B, then A <= B. (I.e., if the end time of A is <= the start time of B, then A<=B.) In the examples, there are two initial streams, representing the schedules of single rooms. In the example, there is not allowed to be any double-booking of a room, so the transmission sequence of the stream is deterministic. However, when the streams are merged, giving Location_Schedule_Harris.json, then there is some overlap in the scheduled periods across the rooms, so the transmission sequence is non-deterministic.