From 833256a45257318409caf24f9b8614dee9813d07 Mon Sep 17 00:00:00 2001
From: daxid <david.rouquet@tetras-libre.fr>
Date: Tue, 18 Jan 2022 10:39:52 +0000
Subject: [PATCH] Add post edit to the OWL extractions to "propagate" the
instances to subclasses
---
unsel_dashboard-iterate.ipynb | 25 ++++++++++++++++++++++---
1 file changed, 22 insertions(+), 3 deletions(-)
diff --git a/unsel_dashboard-iterate.ipynb b/unsel_dashboard-iterate.ipynb
index 7f5e4a4..f473c79 100644
--- a/unsel_dashboard-iterate.ipynb
+++ b/unsel_dashboard-iterate.ipynb
@@ -44,7 +44,7 @@
"extractionGraph = '/opt/dashboards/tetras-lab-unl-demos/work_graph.ttl' # -- old --- extractionGraph = '/opt/dashboards/tetras-lab-unl-demos/extraction-data-9.ttl'\n",
"workDir = 'work-data/'\n",
"webvowlData = '/opt/webvowl'\n",
- "pane_width = 1300\n",
+ "pane_width = 1250\n",
"\n",
"# Fuseki\n",
"fusekiBase = \"https://fuseki.unsel.tetras-lab.io/unsel/\"\n",
@@ -471,7 +471,7 @@
" work_file = dest_file\n",
" inferResult = shaclInfer(work_file, 'infer') # apply SHACL inference\n",
" graph.parse(data=inferResult) # update graph with inference\n",
- " graph.serialize(destination=dest_file, base=base_ref, format='turtle') # serialize graph after inference\n",
+ " #graph.serialize(destination=dest_file, base=base_ref, format='turtle') # serialize graph after inference\n",
" return graph, inferResult"
]
},
@@ -709,6 +709,22 @@
"sparqlUpdate.setQuery('INSERT DATA {'+g.serialize(format='nt')+'}')\n",
"sparqlLog = sparqlUpdate.query()\n",
"\n",
+ "# Requête de post-traitement pour propager les instances aux sous classes pour chaque phrase\n",
+ "# -> Il faudra sans doute revoir la classe qui accueille les instances à l'extraction\n",
+ "def instToSubclasses():\n",
+ " instToSubclassesQuery = \"\"\"PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\n",
+ "PREFIX sys: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/>\n",
+ "INSERT {?inst a ?subClass .}\n",
+ "WHERE { \n",
+ " ?class rdfs:subClassOf* sys:Structure ; sys:from_structure ?sentence .\n",
+ " ?subClass rdfs:subClassOf+ ?class ; sys:from_structure ?sentence . \n",
+ " ?inst a ?class ; sys:from_structure ?sentence . \n",
+ "}\"\"\"\n",
+ " sparqlUpdate.setQuery(instToSubclassesQuery)\n",
+ " sparqlLog = sparqlUpdate.query()\n",
+ "\n",
+ "instToSubclasses()\n",
+ " \n",
"#factoidList = [x.replace('..','/opt/dashboards') for x in glob(storeBase+\"/*/current/*factoid.ttl\")]\n",
"#owl2vowl(FRAME_DIR + frame_file, importList=factoidList)\n",
"\n",
@@ -755,6 +771,7 @@
" factoidGraph.parse(sentencePath) \n",
" sparqlUpdate.setQuery('INSERT DATA {'+factoidGraph.serialize(format='nt')+'}')\n",
" sparqlLog = sparqlUpdate.query()\n",
+ " instToSubclasses()\n",
" createVerificationMessages(storeBase+uuidStr)"
]
},
@@ -770,7 +787,8 @@
"# Fonctions pour la mise à jour globale du corpus (ne pas exécuter en mode \"Tableau de bord\")\n",
"\n",
"def updateAllFactoids():\n",
- " dirList = glob('/opt/dashboards/store/CCTP-SRSA-IP-20210831/*')\n",
+ " dirList = ['/opt/dashboards/store/CCTP-SRSA-IP-20210831/SRSA-IP_STB_PHON_00100']\n",
+ " #dirList = glob('/opt/dashboards/store/CCTP-SRSA-IP-20210831/*')\n",
" for directory in dirList:\n",
" if directory.split('/')[-1] != '0_NONE':\n",
" print(directory)\n",
@@ -781,6 +799,7 @@
" \n",
" \n",
"def updateAllVerificationMessages():\n",
+ " #dirList = ['/opt/dashboards/store/CCTP-SRSA-IP-20210831/SRSA-IP_STB_PHON_00100']\n",
" dirList = glob('/opt/dashboards/store/CCTP-SRSA-IP-20210831/*')\n",
" for directory in dirList:\n",
" if directory.split('/')[-1] != '0_NONE':\n",
--
GitLab