INK rule mining results

ink_results/ink_rulemining/task_agnostic/dbpedia2_ink_complexity_2.log

@@ -0,0 +1,6 @@
+#Process: get neighbourhood
+# Transform
+loading done in:  403.5775247858837
+(1376877, 6946046)
+mining done in: 40191.05462627811
+total time: 40594.632151064

ink_results/ink_rulemining/task_agnostic/dbpedia3.8_ink_complexity_2.log

@@ -0,0 +1,6 @@
+#Process: get neighbourhood
+# Transform
+loading done in:  683.5099277440459
+(2198871, 5647489)
+mining done in: 37982.90089613525
+total time: 38666.4108238793

ink_results/ink_rulemining/task_agnostic/ink_agnostic.py

@@ -0,0 +1,63 @@
+from ink.base.structure import InkExtractor
+from ink.miner.rulemining import RuleSetMiner
+from rdflib_hdt import HDTStore
+from ink.base.connectors import RDFLibConnector, StardogConnector, AbstractConnector
+from sklearn.metrics import accuracy_score
+import sys
+import pickle
+import glob
+from tqdm import tqdm
+from rdflib import URIRef, Graph, Literal
+from timeit import default_timer as timer
+
+class HDTConnector(AbstractConnector):
+
+    def query(self, q_str):
+        global store
+        #if self.all_requests.check(q_str):
+        #    return self.all_requests.get(q_str)
+        try:
+        # res = graph.query(q_str)
+            noi = URIRef(q_str.split('"')[1])
+            res = store.hdt_document.search((noi, None, None))[0]
+            val = [{"p": {"value": r[1].toPython()}, "o": {"value": r[2].n3().split('"')[1]}} if isinstance(r[2],
+                                                                                                            Literal) else {
+                "p": {"value": r[1].toPython()}, "o": {"value": r[2].toPython()}} for r in res]
+            return val
+        # return json.loads(res.serialize(format="json"))#['results']['bindings']
+        except Exception as e:
+            #    print(e)
+            return []
+
+    def get_all_entities(self):
+        global store
+        res = store.hdt_document.search((None, None, None))[0]
+        entities = set()
+        for r in tqdm(res):
+            entities.add(r[0].toPython())
+            #entities.add(r[2].toPython())
+        return entities
+
+
+file = sys.argv[1]
+store = HDTStore(file)
+
+if __name__ == '__main__':    
+    start = timer()
+    connector = HDTConnector()
+    extractor = InkExtractor(connector, verbose=True)
+    pos = connector.get_all_entities()
+
+    X_train, y_train = extractor.create_dataset(2, pos, None, jobs=32)
+    X_train = extractor.fit_transform(X_train)
+    end = timer()
+    print("loading done in: ", str(end-start))
+    model = RuleSetMiner(support=100, max_len_rule_set=3, verbose=True, rule_complexity=int(sys.argv[3]))
+    print(X_train[0].shape)
+
+    rf = model.fit(X_train)
+    end2 = timer()
+    print("mining done in:",str(end2-end))
+    rf.to_csv(sys.argv[2]+'.csv')
+
+    print("total time:",str(end2-start))

ink_results/ink_rulemining/task_agnostic/wikidata_ink_complexity_2.log

@@ -0,0 +1,6 @@
+#Process: get neighbourhood
+# Transform
+loading done in:  776.8928480511531
+(2990435, 2983585)
+mining done in: 28198.46978243813
+total time: 28975.362630489282

ink_results/ink_rulemining/task_agnostic/yago2core_ink_complexity_2.log

@@ -0,0 +1,6 @@
+#Process: get neighbourhood
+# Transform
+loading done in:  47.07918471284211
+(470483, 607102)
+mining done in: 211.44440980697982
+total time: 258.5235945198219

ink_results/ink_rulemining/task_agnostic/yago2s_ink_complexity_2.log

@@ -0,0 +1,6 @@
+#Process: get neighbourhood
+# Transform
+loading done in:  770.2709990260191
+(1653880, 1057539)
+mining done in: 2388.7974461987615
+total time: 3159.0684452247806