- generate_subgraphs(dataset: Dict[str, List], num_subgraphs: int = 5, min_nodes: int = 2, max_nodes: int = 5) List[Dict[str, List[Dict[str, Any]]]]
Generates subgraphs from a given dataset.
This function creates a specified number of subgraphs by randomly selecting nodes and their associated links from the dataset. The size of each subgraph is constrained by the specified minimum and maximum number of nodes.
Parameters
- datasetdict
The input graph dataset containing ‘nodes’ (list of nodes) and ‘links’ (list of edges with ‘source’ and ‘target’ keys).
- num_subgraphsint, optional
The number of subgraphs to generate (default: 5).
- min_nodesint, optional
The minimum number of nodes in each subgraph (default: 2).
- max_nodesint, optional
The maximum number of nodes in each subgraph (default: 5).
Returns
- list
A list of subgraphs, where each subgraph is a dictionary with ‘nodes’ and ‘links’ as keys.
- generate_subgraphs_dataset(subgraphs, large_dataset)
Generate a dataset from a list of subgraphs.
This function takes a list of subgraphs and a large dataset to create smaller datasets corresponding to each subgraph. The smaller datasets contain node features, edge indices, and labels derived from the large dataset, with a mask applied to isolate the subgraph nodes.
- Args:
- subgraphs (list[dict]): A list of subgraphs where each subgraph is represented as a
- dictionary containing:
‘links’ (list[dict]): Edges in the subgraph, where each edge is represented as a dictionary with ‘source’ and ‘target’ node IDs.
‘nodes’ (list[dict]): Nodes in the subgraph, where each node is represented as a dictionary with an ‘id’ key.
- large_dataset (Data): The large dataset containing the following attributes:
x (Tensor): Node feature matrix of shape (num_nodes, num_features).
y (Tensor): Labels for nodes or graphs.
node_mapping (dict): Mapping of node IDs to indices in the dataset.
- Returns:
list[Data]: A list of Data objects, each representing a dataset for a subgraph.