Optimize Similarity using in-memory mutate

Author: JohT

cypher/Similarity/Similarity_1c_Mutate.cypher

@@ -0,0 +1,31 @@
+// Similarity Mutate
+
+ CALL gds.nodeSimilarity.mutate(
+  $dependencies_projection + '-cleaned', {
+      relationshipWeightProperty: $dependencies_projection_weight_property
+     ,topK: 3
+     ,writeRelationshipType: 'SIMILAR'
+     ,writeProperty: 'score'
+ })
+YIELD  preProcessingMillis
+      ,computeMillis
+      ,mutateMillis
+      ,postProcessingMillis
+      ,relationshipsWritten
+      ,nodesCompared
+      ,similarityDistribution
+RETURN preProcessingMillis
+      ,computeMillis
+      ,mutateMillis
+      ,postProcessingMillis
+      ,relationshipsWritten
+      ,nodesCompared
+      ,similarityDistribution.min
+      ,similarityDistribution.mean
+      ,similarityDistribution.max
+      ,similarityDistribution.p50
+      ,similarityDistribution.p75
+      ,similarityDistribution.p90
+      ,similarityDistribution.p95
+      ,similarityDistribution.p99
+      ,similarityDistribution.p999

cypher/Similarity/Similarity_1d_Stream_Mutated.cypher

@@ -0,0 +1,34 @@
+// Read the similarity relationship from the projection. Variables: dependencies_projection
+
+CALL gds.graph.relationships.stream(
+     $dependencies_projection + '-cleaned' 
+    ,['score']
+    ,['SIMILAR']
+)
+YIELD  sourceNodeId
+      ,targetNodeId
+      ,relationshipType
+      ,relationshipProperty
+      ,propertyValue
+// TODO ------------------ Change plain streaming to mutated streaming
+YIELD node1, node2, similarity
+ WITH gds.util.asNode(node1) AS node1
+     ,gds.util.asNode(node2) AS node2
+     ,similarity
+OPTIONAL MATCH (artifact1:Artifact)-[:CONTAINS]->(node1)
+OPTIONAL MATCH (artifact2:Artifact)-[:CONTAINS]->(node2)
+ WITH node1
+     ,node2
+     ,replace(last(split(artifact1.fileName, '/')), '.jar', '') AS artifactName1
+     ,replace(last(split(artifact2.fileName, '/')), '.jar', '') AS artifactName2
+     ,similarity
+RETURN similarity
+      ,artifactName1
+      ,coalesce(node1.fqn, node1.fileName, node1.name) AS node1Name
+      ,node1.incomingDependencies                      AS node1IncomingDependencies  
+      ,node1.outgoingDependencies                      AS node1OutgoingDependencies   
+      ,artifactName2
+      ,coalesce(node2.fqn, node2.fileName, node2.name) AS node2Name
+      ,node2.incomingDependencies                      AS node2IncomingDependencies  
+      ,node2.outgoingDependencies                      AS node2OutgoingDependencies  
+ORDER BY similarity DESCENDING, node1Name, node2Name

cypher/Similarity/Similarity_1c_Stream.cypher renamed to cypher/Similarity/Similarity_1e_Stream.cypher

@@ -0,0 +1,8 @@
+// Write a property from the projection into the Graph. Variables: dependencies_projection, dependencies_projection_write_property
+
+CALL gds.graph.nodeProperties.write(
+   $dependencies_projection + '-cleaned'
+  ,[$dependencies_projection_write_property]
+)
+ YIELD propertiesWritten, nodeProperties, writeMillis
+RETURN propertiesWritten, nodeProperties, writeMillis

cypher/Similarity/Similarity_1d_Delete_Relationships.cypher renamed to cypher/Similarity/Similarity_1f_Delete_Relationships.cypher

@@ -72,15 +72,18 @@ similarity() {
     execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1a_Estimate.cypher" "${@}"
     execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1b_Statistics.cypher" "${@}"
 
+    # Run the algorithm and write the result into the in-memory projection ("mutate")
+    execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1c_Mutate.cypher" "${@}"
+
     # Stream to CSV
     local nodeLabel
     nodeLabel=$( extractQueryParameter "dependencies_projection_node" "${@}" )
-    execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1c_Stream.cypher" "${@}" > "${FULL_REPORT_DIRECTORY}/${nodeLabel}_Similarity.csv"
+    execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1e_Stream.cypher" "${@}" > "${FULL_REPORT_DIRECTORY}/${nodeLabel}_Similarity.csv"
 
     # Update Graph (node properties and labels)
-    execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1d_Delete_Relationships.cypher" "${@}"
-    execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1e_Write.cypher" "${@}"
-    execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1f_Write_Node_Properties.cypher" "${@}"
+    execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1g_Delete_Relationships.cypher" "${@}"
+    execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1h_Write.cypher" "${@}"
+    execute_cypher "${SIMILARITY_CYPHER_DIR}/Similarity_1i_Write_Node_Properties.cypher" "${@}"
 }
 
 # ---------------------------------------------------------------