From a4b9dae35255df7380b984f22d53b5e9f3d4a491 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Aur=C3=A9lien=20Lamercerie?=
<aurelien.lamercerie@tetras-libre.fr>
Date: Thu, 21 Sep 2023 17:55:30 +0200
Subject: [PATCH] Preprocessing AMR Rule to fix some bug on SSC-04 graph
---
tenet/scheme/amr_master_rule/__init__.py | 2 +
.../preprocessing/amr_bug_fixing_2.py | 88 +++++++++++++++++
.../preprocessing/amr_bug_fixing_3.py | 94 +++++++++++++++++++
tenet/scheme/owl_amr_scheme_1.py | 4 +-
tests/main_tests/test_main_owl_extraction.py | 2 +-
5 files changed, 188 insertions(+), 2 deletions(-)
create mode 100644 tenet/scheme/amr_master_rule/preprocessing/amr_bug_fixing_2.py
create mode 100644 tenet/scheme/amr_master_rule/preprocessing/amr_bug_fixing_3.py
diff --git a/tenet/scheme/amr_master_rule/__init__.py b/tenet/scheme/amr_master_rule/__init__.py
index da108ca3..ef2bd7ce 100644
--- a/tenet/scheme/amr_master_rule/__init__.py
+++ b/tenet/scheme/amr_master_rule/__init__.py
@@ -5,6 +5,8 @@
# -- Preprocessing Rules
from scheme.amr_master_rule.preprocessing.amr_bug_fixing_1 import *
+from scheme.amr_master_rule.preprocessing.amr_bug_fixing_2 import *
+from scheme.amr_master_rule.preprocessing.amr_bug_fixing_3 import *
from scheme.amr_master_rule.preprocessing.amr_reification_1_concept import *
from scheme.amr_master_rule.preprocessing.amr_reification_2_concept import *
diff --git a/tenet/scheme/amr_master_rule/preprocessing/amr_bug_fixing_2.py b/tenet/scheme/amr_master_rule/preprocessing/amr_bug_fixing_2.py
new file mode 100644
index 00000000..929d0614
--- /dev/null
+++ b/tenet/scheme/amr_master_rule/preprocessing/amr_bug_fixing_2.py
@@ -0,0 +1,88 @@
+#!/usr/bin/python3.10
+# -*- coding: Utf-8 -*-
+
+#==============================================================================
+# TENET: Rule to fix AMR bug
+#------------------------------------------------------------------------------
+# Fix AMR bug about planet named system solar
+#==============================================================================
+
+from rdflib import Graph, Literal
+
+import transduction
+from transduction import net
+from transduction.query_builder import generate_select_query
+from transduction.rdfterm_computer import ( produce_uriref, produce_literal )
+
+
+#==============================================================================
+# Pattern Search: bug about planet named system solar
+#==============================================================================
+
+def __search_pattern(graph):
+ select_data_list = ['?originNode', '?originConcept']
+ clause_list = ['?originConcept a ns3:Concept.',
+ 'FILTER ( CONTAINS(str(?originConcept), str(ns2:intercontinental)) ).',
+ '?originNode a ?originConcept.',
+ '?cloudNode ns2:mod ?originNode.',
+ '?cloudNode a ns2:cloud.'
+ ]
+ query_code = generate_select_query(graph, select_data_list, clause_list)
+ result_set = graph.query(query_code)
+ return query_code, result_set
+
+
+#==============================================================================
+# Construct Method(s)
+#==============================================================================
+
+def __add_new_concept(graph, originNode, originConcept, new_label='interstellar'):
+ triple_list = []
+
+ new_concept_uri = produce_uriref(graph, f'ns2:{new_label}')
+
+ # -- New concept
+ relation = produce_uriref(graph, 'rdf:type')
+ value = produce_uriref(graph, 'ns3:Concept')
+ triple_list.append((new_concept_uri, relation, value))
+
+ # # -- Label
+ # relation = produce_uriref(graph, 'rdfs:label')
+ # value = Literal(new_label)
+ # triple_list.append((new_concept_uri, relation, value))
+
+ # -- Linking between individual and new concept
+ relation = produce_uriref(graph, 'rdf:type')
+ triple_list.append((originNode, relation, new_concept_uri))
+
+ # -- Classification of the original concept as LinkedData (for tracing)
+ relation = produce_uriref(graph, 'rdfs:subClassOf')
+ value = produce_uriref(graph, 'amr:AMR_Linked_Data')
+ triple_list.append((originConcept, relation, value))
+
+ # -- Comment the origin concept as 'bug'
+ relation = produce_uriref(graph, 'rdfs:comment')
+ value = Literal('bug')
+ triple_list.append((originConcept, relation, value))
+
+ return triple_list
+
+
+#==============================================================================
+# Main Method
+#==============================================================================
+
+def fix_amr_bug_2(graph):
+
+ # -- Rule Initialization
+ rule_label = 'fix AMR bug (2)'
+
+ # -- Search for patterns
+ _, pattern_set = __search_pattern(graph)
+
+ # -- Selection Analyzing (1)
+ rule_triple_list = []
+ for pattern in pattern_set:
+ rule_triple_list += __add_new_concept(graph, pattern.originNode, pattern.originConcept)
+
+ return rule_label, rule_triple_list
diff --git a/tenet/scheme/amr_master_rule/preprocessing/amr_bug_fixing_3.py b/tenet/scheme/amr_master_rule/preprocessing/amr_bug_fixing_3.py
new file mode 100644
index 00000000..5849b4fa
--- /dev/null
+++ b/tenet/scheme/amr_master_rule/preprocessing/amr_bug_fixing_3.py
@@ -0,0 +1,94 @@
+#!/usr/bin/python3.10
+# -*- coding: Utf-8 -*-
+
+#==============================================================================
+# TENET: Rule to fix AMR bug
+#------------------------------------------------------------------------------
+# Fix AMR bug about planet named system solar
+#==============================================================================
+
+from rdflib import Graph, Literal
+
+import transduction
+from transduction import net
+from transduction.query_builder import generate_select_query
+from transduction.rdfterm_computer import ( produce_uriref, produce_literal )
+
+
+#==============================================================================
+# Pattern Search: bug about planet named system solar
+#==============================================================================
+
+def __search_pattern(graph):
+ select_data_list = ['?originNode', '?originConcept', '?cloudNode']
+ clause_list = ['?originConcept a ns3:Concept.',
+ 'FILTER ( CONTAINS(str(?originConcept), str(ns2:molecule)) ).',
+ '?originNode a ?originConcept.',
+ '?originNode ns2:consist ?cloudNode.',
+ '?cloudNode a ns2:cloud.'
+ ]
+ query_code = generate_select_query(graph, select_data_list, clause_list)
+ result_set = graph.query(query_code)
+ return query_code, result_set
+
+
+#==============================================================================
+# Construct Method(s)
+#==============================================================================
+
+def __add_new_concept(
+ graph, originNode, originConcept, cloudNode, new_label='molecular'):
+ triple_list = []
+
+ new_concept_uri = produce_uriref(graph, f'ns2:{new_label}')
+
+ # -- New concept
+ relation = produce_uriref(graph, 'rdf:type')
+ value = produce_uriref(graph, 'ns3:Concept')
+ triple_list.append((new_concept_uri, relation, value))
+
+ # # -- Label
+ # relation = produce_uriref(graph, 'rdfs:label')
+ # value = Literal(new_label)
+ # triple_list.append((new_concept_uri, relation, value))
+
+ # -- Linking between individual and new concept
+ relation = produce_uriref(graph, 'rdf:type')
+ triple_list.append((originNode, relation, new_concept_uri))
+
+ # -- Linking between individual and linked node
+ relation = produce_uriref(graph, 'ns2:mod')
+ triple_list.append((cloudNode, relation, originNode))
+
+ # -- Classification of the original concept as LinkedData (for tracing)
+ relation = produce_uriref(graph, 'rdfs:subClassOf')
+ value = produce_uriref(graph, 'amr:AMR_Linked_Data')
+ triple_list.append((originConcept, relation, value))
+
+ # -- Comment the origin concept as 'bug'
+ relation = produce_uriref(graph, 'rdfs:comment')
+ value = Literal('bug')
+ triple_list.append((originConcept, relation, value))
+
+ return triple_list
+
+
+#==============================================================================
+# Main Method
+#==============================================================================
+
+def fix_amr_bug_3(graph):
+
+ # -- Rule Initialization
+ rule_label = 'fix AMR bug (3)'
+
+ # -- Search for patterns
+ _, pattern_set = __search_pattern(graph)
+
+ # -- Selection Analyzing (1)
+ rule_triple_list = []
+ for pattern in pattern_set:
+ rule_triple_list += __add_new_concept(
+ graph, pattern.originNode, pattern.originConcept, pattern.cloudNode)
+
+ return rule_label, rule_triple_list
diff --git a/tenet/scheme/owl_amr_scheme_1.py b/tenet/scheme/owl_amr_scheme_1.py
index d461b2e9..5a8e2f02 100644
--- a/tenet/scheme/owl_amr_scheme_1.py
+++ b/tenet/scheme/owl_amr_scheme_1.py
@@ -48,7 +48,9 @@ prefix_list = [('owl', '<http://www.w3.org/2002/07/owl#>'),
# ---------------------------------------------
amr_bug_fixing_sequence = ['Bug fixing for some known anomalies of AMR-LD data',
- rule.fix_amr_bug_1
+ rule.fix_amr_bug_1,
+ rule.fix_amr_bug_2,
+ rule.fix_amr_bug_3
]
amr_reification_sequence = ['AMR reification from AMR-Linked-Data to AMR (tenet) structure',
diff --git a/tests/main_tests/test_main_owl_extraction.py b/tests/main_tests/test_main_owl_extraction.py
index a89785c4..ef595bb1 100644
--- a/tests/main_tests/test_main_owl_extraction.py
+++ b/tests/main_tests/test_main_owl_extraction.py
@@ -56,7 +56,7 @@ test_data_dir = f'{INPUT_DIR_PATH}amrDocuments/'
# onto_prefix = f'SimpleTest'
# base_output_name = f'SimpleTest'
-uuid_num = '02'
+uuid_num = '04'
amrld_dir_path = f'{test_data_dir}dev/solar-system-{uuid_num}/'
amrld_file_path = f'{amrld_dir_path}SSC-{uuid_num}-01.stog.amr.ttl'
base_output_name = f'SolarSystemDev{uuid_num}'
--
GitLab