Using pyroscope I see that about 15% of the time is used for determining and extracting prefixes using regex on predicates and object URIs while prefixes are only applicable when returning Turtle format.
Even more, this results in ugly and strange prefixes, while it suffices to use well-known prefixes. However, it would be useful to also add a few additional prefixes as most objects in the LDES are prefixed with the LDES server URL and optionally allow to configure a set of prefixes (system wide).
Suggestion, for Turtle format do not determine/extract prefixes but simply always add:
a set of well-known prefixes (rdfs:, dct:, xml:, etc.)
a set of LDES specific prefixes, e.g. {ldes-name}:, {view-name}:, etc.
Using pyroscope I see that about 15% of the time is used for determining and extracting prefixes using regex on predicates and object URIs while prefixes are only applicable when returning Turtle format.
Even more, this results in ugly and strange prefixes, while it suffices to use well-known prefixes. However, it would be useful to also add a few additional prefixes as most objects in the LDES are prefixed with the LDES server URL and optionally allow to configure a set of prefixes (system wide).
Suggestion, for Turtle format do not determine/extract prefixes but simply always add:
rdfs:,dct:,xml:, etc.){ldes-name}:,{view-name}:, etc.