OWL TIME functions implemented as SPARQL extension functions in rdflib
Project description
RDFlib OWL TIME Functions
This Python package implements functions to ascertain the temporal relations of objects in RDF graphs.
This package invents SPARQL functions that are based on logic provided in OWL TIME and are named similarly to TIME's Temporal Entity's predicates, for example, the function isBefore()
correlates to time:before
, i.e. isBefore(a, b)
will test to see if time:before
should be applied between object a
& b
.
The functions are made available as functions in RDFlib's SPARQL implementation where they can be called by IRI, e.g. isBefore(a, b)
is called tfun:isBefore(?a, ?b)
. All functions are posed as questions, "is before", and return an RDF literal true
or false
.
This repository also contains a formal declaration of the functions that this package implements as a SKOS vocabulary. See the Vocabulary section below.
Installation
Normal installation e.g. from PyPI: pip install timefuncs
.
This package's only dependency is RDFlib.
Use
These functions are implemented in RDFlib Python in the file timefuncts/funcs.py
and are imported into timefuncs/__init__.py
and registered there in RDFlib as SPARQL extension functions with their IRIs.
This means they can be used like this (full working script):
from rdflib import Graph
from timefuncs import TFUN
data = """
PREFIX : <http://example.com/>
PREFIX time: <http://www.w3.org/2006/time#>
:a01 a time:TemporalEntity .
:b01 a time:TemporalEntity .
:c01 a time:TemporalEntity .
:d01 a time:TemporalEntity .
:a01 time:before :c01 .
:b01 time:after :c01 .
"""
g = Graph().parse(data=data)
q = """
PREFIX tfun: <https://w3id.org/timefuncs/>
SELECT *
WHERE {
?x a time:TemporalEntity .
?y a time:TemporalEntity .
FILTER tfun:isBefore(?x, ?y)
}
"""
for r in g.query(q):
print(f"{r['x']} is before {r['y']}")
The above stript outputs:
http://example.com/a01 is before http://example.com/b01
http://example.com/a01 is before http://example.com/c01
http://example.com/b01 is before http://example.com/c01
The above script is run using an environment that has had the time functions registered with its copy of rdflib (perhaps by running pip install timefuncs
) so that there is no need to import anything other than rdfib and the time functions namespace (from timefuncs import TFUN
). It may appear that the namespace is not used but it is, internally! No need to re-declare tfun:
as a PREFIX
in the SPARQL query...
The time function used here, tfun:isBefore
, is called as a filter function to return true
when the first given object, here ?x
is before the second given object, ?y
.
This example uses a pretty open-ended graph pattern match (`?x ?p ?)
Functions
Functions in this package are implemented as SPARQL extension functions with the namespace https://w3id.org/timefuncs/
, e.g. isBefore()
's full IRI is https://w3id.org/timefuncs/isBefore
.
Functions implemented so far, and their corresponding TIME relations:
SPARQL | Python | TIME predicates | Notes |
---|---|---|---|
tfun:contains() |
contains() |
time:intervalContains inv. time:intervalDuring |
|
tfun:hasDuring() |
has_during() |
alias for contains() |
|
tfun:hasInside() |
has_inside() |
time:inside not time:before not time:after |
|
tfun:isAfter() |
is_after() |
time:after inv. time:before |
|
tfun:isBefore() |
is_before() |
time:before inv. time:after |
|
tfun:isContainedBy() |
is_contained_by() |
time:intervalDuring inv. time:intervalContains |
|
tfun:isDuring() |
is_during() |
alias for isContainedBy() |
|
tfun:isInside() |
is_inside() |
inv. time:inside not time:after not time:before |
These functions are yet to be implemented:
SPARQL | Python | TIME predicates | Notes |
---|---|---|---|
tfun:finishes() |
finishes() |
||
tfun:hasBeginning() |
has_beginning() |
||
tfun:hasEnd() |
has_end() |
||
tfun:isFinishedBy() |
is_finished_by() |
||
tfun:isBeginningOf() |
is_beginning_of() |
||
tfun:isDisjoint() |
is_disjoint() |
||
tfun:isEndOf() |
is_end_of() |
||
tfun:isEquals() |
is_equals() |
||
tfun:isIn() |
is_in() |
||
tfun:isMetBy() |
is_met_by() |
||
tfun:isNotDisjoint() |
is_not_disjoint() |
||
tfun:isOverlappedBy() |
is_overlapped_by() |
||
tfun:isStartedBy() |
is_started_by() |
||
tfun:meets() |
meets() |
||
tfun:overlaps() |
overlaps() |
||
tfun:starts() |
starts() |
Implementation logic
Functions implemented test for every conceivable way that a temporal relation may be found to be true in given data. For example, isBefore(a, b)
will return true if:
- the predicate
time:before
is given betweena
&b
, or any chain of objectsa
...n
...b
- the predicate
time:after
is given betweenb
&a
, or any chain of objectsb
...n
...a
- any chain of
time:before
&time:after
linksa
&b
, all pointing in the right directions - the end of
a
is declared as being beforeb
or the start ofb
- i.e. for
<a> time:hasEnd <a_end> .
and<b> time:hasBeginning <b_beginning> .
,isBefore(a, b)
istrue
if<a_end> time:before <b_beginning>
- i.e. for
- the start of
b
is declared as being aftera
or the end ofa
a
can be calculated as being beforeb
, based on their instantaneous times or start and end times- i.e. for
<a> time:inXSDDateTimeStamp <a_xsd> .
and<b> time:inXSDDateTimeStamp <b_xsd> .
or<a> time:hasEnd/time:inXSDDateTimeStamp <a_xsd> .
and<b> time:hasEnd/time:inXSDDateTimeStamp <b_xsd> .
,isBefore(a, b)
istrue
if<a_xsd> <b_xsd>
- i.e. for
Vocabulary
The time functions, both implemented and to-be implemented, are listed in a SKOS vocabulary, the source files for which are given in the voc/ folder within this repository. The vocabulary is presented online in both RDF (turtle) and HTML (Markdown) formats from these source files, accessible via the namespace IRI:
Each function's IRI resolves to its Concept entry in the vocbaulary, for example, https://w3id.org/timefuncs/isContainedBy --> https://github.com/RDFLib/timefuncs/blob/master/voc/timefuncs.md#is-contained-by.
(The Markdown is auto-generated from the RDF using pyLODE)
Testing
All tests are in tests/
and implemented using pytest.
There are individual tests for each function, e.g. tests/test_is_before.py
for isBefore()
as well as a test file to rn all tests againts OWL TIME's test suite): tests/test_test_suite.py
.
Contributing
Via GitHub, Issues & Pull Requests:
License
This code is licensed with the BSD 3-clause license as per LICENSE which is the same license as used for rdflib.
Citation
@software{https://github.com/rdflib/timefuncs,
author = {{Nicholas J. Car}},
title = {RDFlib OWL TIME Functions},
version = {0.0.2},
date = {2021},
url = {https://github.com/rdflib/timefuncs}
}
Contact
Creator & maintainer:
Nicholas J. Car
Data System Architect
SURROUND Australia Pty Ltd
nicholas.car@surroundaustrlaia.com
and
Adjunct Senior Lecturer
College of Engineering & Computer Science
Australian National University
nicholas.car@anu.edu.au
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for timefuncs-0.1.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 33fd1201665f0328ac5bc687a0d5ffbc3b9416c91ce373ea909873f0db9ca03f |
|
MD5 | 5247f7682a70bb129328f86421391403 |
|
BLAKE2b-256 | 5e4d0bdc41c83f8dadf315417eb07fbb7df4c2daa57be0376f46653a17f6fe3a |