8.2.4 Building a GraphWise Model Using Heterogeneous Graphs
You can build a GraphWise model using partitioned graphs which have different providers and features.
opg4j> analyst.supervisedGraphWiseModelBuilder().
setVertexInputPropertyNames("vertex_provider1_features", "vertex_provider2_features").
setEdgeInputPropertyNames("edge_provider_features").
setVertexTargetPropertyName("target_property").
build()
SupervisedGraphWiseModel model = analyst.supervisedGraphWiseModelBuilder()
.setVertexInputPropertyNames("vertex_provider1_features", "vertex_provider2_features")
.setEdgeInputPropertyNames("edge_provider_features")
.setVertexTargetPropertyName("target_property")
.build();
params = dict(vertex_target_property_name="target_property",
vertex_input_property_names=["vertex_provider1_features", "vertex_provider2_features"],
edge_input_property_names=["edge_provider_features"])
model = analyst.supervised_graphwise_builder(**params)
Also, you can select the providers as shown:
opg4j> var model = analyst.supervisedGraphWiseModelBuilder().
setVertexInputPropertyNames("vertex_provider1_features", "vertex_provider2_features").
setEdgeInputPropertyNames("edge_provider_features").
setVertexTargetPropertyName("target_property").
setTargetVertexLabels("provider1").
build()
SupervisedGraphWiseModel model = analyst.supervisedGraphWiseModelBuilder()
.setVertexInputPropertyNames("vertex_provider1_features", "vertex_provider2_features")
.setEdgeInputPropertyNames("edge_provider_features")
.setVertexTargetPropertyName("target_property")
.setTargetVertexLabels("provider1")
.build();
params = dict(vertex_target_property_name="target_property",
vertex_input_property_names=["vertex_provider1_features", "vertex_provider2_features"],
edge_input_property_names=["edge_provider_features"],
target_vertex_labels=["provider1"])
model = analyst.supervised_graphwise_builder(**params)
If you wish to control the flow of the embeddings at each layer, you can enable or disable the required connections. By default, all the connections are enabled.
opg4j> var convLayerConfig = analyst.graphWiseConvLayerConfigBuilder().
setNumSampledNeighbors(25).
useVertexToVertexConnection(true).
useEdgeToVertexConnection(true).
useEdgeToEdgeConnection(false).
useVertexToEdgeConnection(false).
build()
opg4j> var model = analyst.supervisedGraphWiseModelBuilder().
setVertexInputPropertyNames("vertex_provider1_features", "vertex_provider2_features").
setEdgeInputPropertyNames("edge_provider_features").
setVertexTargetPropertyName("target_property").
setTargetVertexLabels("provider1").
build()
GraphWiseConvLayerConfig convLayerConfig = analyst.graphWiseConvLayerConfigBuilder()
.setNumSampledNeighbors(10)
.useVertexToVertexConnection(true)
.useEdgeToVertexConnection(true)
.useEdgeToEdgeConnection(false)
.useVertexToEdgeConnection(false)
.build();
SupervisedGraphWiseModel model = analyst.supervisedGraphWiseModelBuilder()
.setVertexInputPropertyNames("vertex_provider1_features", "vertex_provider2_features")
.setEdgeInputPropertyNames("edge_provider_features")
.setVertexTargetPropertyName("target_property")
.setTargetVertexLabels("provider1")
.setConvLayerConfigs(convLayerConfig)
.build();
conv_layer_config = dict(num_sampled_neighbors=25,
activation_fn='tanh',
weight_init_scheme='xavier',
neighbor_weight_property_name=weightProperty,
vertex_to_vertex_connection=True,
edge_to_vertex_connection=True,
vertex_to_edge_connection=False,
edge_to_edge_connection=False)
conv_layer = analyst.graphwise_conv_layer_config(**conv_layer_config)
params = dict(vertex_target_property_name="target_property",
vertex_input_property_names=["vertex_provider1_features", "vertex_provider2_features"],
edge_input_property_names=["edge_provider_features"],
target_vertex_labels=["provider1"],
conv_layer_config=[conv_layer])
model = analyst.supervised_graphwise_builder(**params)
Parent topic: Using the Supervised GraphWise Algorithm