Adding optimized SHACL shapes to reduce inference time

Makefile

@@ -19,7 +19,7 @@ format:
 	black tools/
 
 test: Brick.ttl shacl/BrickShape.ttl
-	pytest -s -vvvv -m 'not slow' tests
+	pytest -s -vvvv -m 'not slow' --durations=0 tests 
 	cd tests/integration && bash run_integration_tests.sh
 
 quantity-test: Brick.ttl

bricksrc/equipment.py

@@ -186,7 +186,7 @@
         },
     },
     "Fire_Safety_Equipment": {
-        "tags": [TAG.Equipment, TAG.Fire, TAG.Safety, TAG.Equipment],
+        "tags": [TAG.Equipment, TAG.Fire, TAG.Safety],
         "subclasses": {
             "Fire_Control_Panel": {
                 "tags": [TAG.Equipment, TAG.Fire, TAG.Safety, TAG.Panel],

generate_brick.py

@@ -189,22 +189,29 @@ def add_tags(klass, definition):
     num_tags = len(definition)
     if len(definition) not in has_exactly_n_tags_shapes:
         # tag count condition
-        cond = BNode(f"has_exactly_{num_tags}_tags_condition")
+        cond = BSH[f"has_exactly_{num_tags}_tags_condition"]
         prop = BNode(f"has_exactly_{num_tags}_tags")
+        shaclGraph.add((cond, A, OWL.Class))
+        shaclGraph.add((cond, A, SH.NodeShape))
         shaclGraph.add((cond, SH.property, prop))
         shaclGraph.add((prop, SH.path, BRICK.hasTag))
         shaclGraph.add((prop, SH.minCount, Literal(len(definition))))
         shaclGraph.add((prop, SH.maxCount, Literal(len(definition))))
         has_exactly_n_tags_shapes[len(definition)] = cond
-    shaclGraph.add((rule, SH.condition, has_exactly_n_tags_shapes[len(definition)]))
-
-    # ensure that the rule applies to at least one of the base tags that should be on
-    # most Brick classes
-    # base_tags = [TAG.Equipment, TAG.Point, TAG.Location, TAG.System, TAG.Solid, TAG.Fluid]
-    # target_class_tag = [t for t in base_tags if t in definition]
-    # assert len(target_class_tag) > 0, klass
-    # shaclGraph.add((sc, SH.targetClass, has_tag_restriction_class[target_class_tag[0]]))
-    shaclGraph.add((sc, SH.targetSubjectsOf, BRICK.hasTag))
+
+        # generate inference rule
+        rule = BSH[f"has_exactly_{num_tags}_tags_rule"]
+        body = BNode(f"has_{num_tags}_tags_body")
+        shaclGraph.add((rule, A, SH.NodeShape))
+        shaclGraph.add((rule, SH.targetSubjectsOf, BRICK.hasTag))
+        shaclGraph.add((rule, SH.rule, body))
+        shaclGraph.add((body, A, SH.TripleRule))
+        shaclGraph.add((body, SH.subject, SH.this))
+        shaclGraph.add((body, SH.predicate, RDF.type))
+        shaclGraph.add((body, SH.object, cond))
+        shaclGraph.add((body, SH.condition, cond))
+
+    shaclGraph.add((sc, SH.targetClass, has_exactly_n_tags_shapes[len(definition)]))
 
     # if we've already mapped this class, don't map it again
     if klass in intersection_classes:

requirements.txt

@@ -6,9 +6,9 @@ rdflib-jsonld==0.5.0
 six==1.12.0
 pytest>=5.4.3
 tqdm>=4.46.1
-pyshacl>=0.15.0
+pyshacl>=0.16.1.post1
 docker>=4.3.0
-brickschema>=0.5.0a1
+brickschema>=0.5.1a1
 black==21.5b2
 pre-commit>=2.13.0
 flake8==3.9.2

tests/conftest.py

@@ -16,3 +16,8 @@ def pytest_generate_tests(metafunc):
         example_files = glob.glob("examples/*/*.ttl")
         # example_files = set(example_files_1 + example_files_2)
         metafunc.parametrize("filename", example_files)
+
+def pytest_configure(config):
+    config.addinivalue_line(
+            "markers", "slow: mark tests as slow (deselect w/ '-m \"not slow\"')"
+            )

tests/test_hierarchy_inference.py

@@ -96,11 +96,16 @@ def test_hierarchyinference():
         )  # This is based on how the entity name is defined above.
 
         # Find the original classes through the hierarchy from the original graph.
-        qstr = """
-        select ?parent where {{
-            <{0}> rdfs:subClassOf* ?parent.
-            ?parent rdfs:subClassOf* brick:Class.
-        }}""".format(
+        qstr = """select ?parent where {{
+            {{
+                <{0}> owl:equivalentClass*/rdfs:subClassOf*/owl:equivalentClass*/rdfs:subClassOf* ?parent .
+            }} UNION {{
+                ?other owl:equivalentClass* <{0}> .
+                ?other rdfs:subClassOf*/owl:equivalentClass*/rdfs:subClassOf* ?parent .
+            }}
+            ?parent rdfs:subClassOf* brick:Class
+        }}
+        """.format(
             true_class
         )
         res = g.query(qstr)

tests/test_inference.py

@@ -66,7 +66,7 @@ def test_tag_inference():
 
     # Apply reasoner
     g.load_file("extensions/brick_extension_shacl_tag_inference.ttl")
-    g.expand(profile="owlrl+shacl+owlrl")
+    g.expand(profile="shacl+owlrl+shacl+owlrl")
 
     g.bind("rdf", RDF)
     g.bind("owl", OWL)

tests/test_measures_inference.py

@@ -66,7 +66,7 @@ def test_measures_infers():
 
     # Infer classes of the entities.
     # Apply reasoner
-    g.expand(profile="owlrl")
+    g.expand(profile="brick")
 
     qstr = """select ?instance ?class where {
         ?instance a ?class.
@@ -94,8 +94,13 @@ def test_measures_infers():
 
         # Find the original classes through the hierarchy from the original graph.
         qstr = """select ?parent where {{
-            <{0}> rdfs:subClassOf* ?parent.
-            ?parent rdfs:subClassOf* brick:Class.
+            {{
+                <{0}> owl:equivalentClass*/rdfs:subClassOf*/owl:equivalentClass*/rdfs:subClassOf* ?parent .
+            }} UNION {{
+                ?other owl:equivalentClass* <{0}> .
+                ?other rdfs:subClassOf*/owl:equivalentClass*/rdfs:subClassOf* ?parent .
+            }}
+            ?parent rdfs:subClassOf* brick:Class
         }}
         """.format(
             true_class