add confidence for most likely host software of plugins

f-krueger · Dec 14, 2021 · f4bcba2 · f4bcba2
1 parent 730e4f6
commit f4bcba2
Showing 1 changed file with 127 additions and 3 deletions.
diff --git a/Creation/02_SoftwareKG_Inference.ipynb b/Creation/02_SoftwareKG_Inference.ipynb
@@ -80,7 +80,8 @@
     "        \"dct\" : \"http://purl.org/dc/terms/\",\n",
     "        \"dbpedia-owl\" : \"http://dbpedia.org/ontology/\",\n",
     "        \"skgv\" : \"http://data.gesis.org/softwarekg/vocab/\",\n",
-    "        \"skg\" : \"http://data.gesis.org/softwarekg/PMC/\"\n",
+    "        \"skg\" : \"http://data.gesis.org/softwarekg/PMC/\",\n",
+    "        \"irao\" : \"http://ontology.ethereal.cz/irao/\"\n",
     "}"
    ]
   },
@@ -419,7 +420,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# Host software of plugins"
+    "# PlugIns of host software"
    ]
   },
   {
@@ -516,7 +517,7 @@
    "source": [
     "# Add statements with confidence value\n",
     "for idx, r in enumerate(plugin_table):\n",
-    "    node = URIRef(\"skg:inference/plugInOf/{}\".format(idx))\n",
+    "    node = URIRef(\"skg:inference/hasPlugIn/{}\".format(idx))\n",
     "    g.add((node, RDF.type, RDF.Statement))\n",
     "    g.add((node, RDF.subject, URIRef(r[0])))\n",
     "    g.add((node, RDF.predicate ,URIRef(\"schema:softwareAddOn\")))\n",
@@ -537,6 +538,129 @@
     "g = Graph()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Host software of plugIns"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "q1='''\n",
+    "PREFIX schema: <http://schema.org/>\n",
+    "PREFIX nif: <http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#>\n",
+    "PREFIX skgv: <http://data.gesis.org/softwarekg/vocab/>\n",
+    "\n",
+    "SELECT ?document ?plugIn ?hostSoftware (COUNT(?hostSoftware) as ?num_mention_per_software_per_plugin_per_document)\n",
+    "WHERE\n",
+    "{\n",
+    "    ?hostMention skgv:software ?hostSoftware .\n",
+    "    ?hostMention skgv:referredToByPlugIn ?plugInMention .\n",
+    "    ?plugInMention skgv:software ?plugIn .\n",
+    "    ?document schema:mentions ?plugInMention .\n",
+    "}\n",
+    "GROUP BY ?document ?plugIn ?hostSoftware\n",
+    "OFFSET {offset}\n",
+    "LIMIT {limit}\n",
+    "'''\n",
+    "table1 = build_table(q1, 1_000_000, ['document','plugIn','hostSoftware',\n",
+    "                                     'num_mention_per_software_per_plugin_per_document'])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "q2=\"\"\"\n",
+    "PREFIX schema: <http://schema.org/>\n",
+    "PREFIX nif: <http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#>\n",
+    "PREFIX skgv: <http://data.gesis.org/softwarekg/vocab/>\n",
+    "\n",
+    "SELECT ?document ?plugIn (COUNT(?plugIn) AS ?num_mention_per_plugin_per_document)\n",
+    "WHERE\n",
+    "{\n",
+    "    ?plugInMention skgv:software ?plugIn .\n",
+    "    ?anyHostMention skgv:referredToByPlugIn ?plugInMention .\n",
+    "    ?document schema:mentions ?plugInMention .\n",
+    "}\n",
+    "GROUP BY ?document ?plugIn\n",
+    "OFFSET {offset}\n",
+    "LIMIT {limit}\n",
+    "\"\"\"\n",
+    "table2 = build_table(q2, 1_000_000, ['document','plugIn', 'num_mention_per_plugin_per_document'])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "q3=\"\"\"\n",
+    "PREFIX schema: <http://schema.org/>\n",
+    "PREFIX nif: <http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#>\n",
+    "PREFIX skgv: <http://data.gesis.org/softwarekg/vocab/>\n",
+    "\n",
+    "SELECT COUNT(DISTINCT ?document) as ?num_article_with_plugin ?plugIn \n",
+    "WHERE\n",
+    "{\n",
+    "    ?plugInMention skgv:software ?plugIn .\n",
+    "    ?someHostMention skgv:referredToByPlugIn ?plugInMention .\n",
+    "    ?document schema:mentions ?plugInMention .\n",
+    "}\n",
+    "GROUP BY  ?plugIn\n",
+    "OFFSET {offset}\n",
+    "LIMIT {limit}\n",
+    "\"\"\"\n",
+    "table3 = build_table(q3, 1_000_000, ['plugIn','num_article_with_plugin'])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "host_table = workflow(table1,table2,table3)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Add statements with confidence value\n",
+    "g = Graph()\n",
+    "for idx, r in enumerate(host_table):\n",
+    "    node = URIRef(\"skg:inference/hasHostSoftware/{}\".format(idx))\n",
+    "    g.add((node, RDF.type, RDF.Statement))\n",
+    "    g.add((node, RDF.subject, URIRef(r[0])))\n",
+    "    g.add((node, RDF.predicate ,URIRef(\"irao:isPartOfSystem\")))\n",
+    "    g.add((node, RDF.object, URIRef(r[1])))\n",
+    "    g.add((node, URIRef(\"skgv:confidence\"), Literal(r[2], datatype=XSD.float)))\n",
+    "    if float(r[2]) < 1:\n",
+    "        print(r)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Serialize\n",
+    "g.serialize(format=\"json-ld\", context=context, destination=\"softwarekg-hostsoftware-confidence.jsonld\")\n",
+    "g = Graph()"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},