Module BML.transform.temporalGraph

Classes

class TemporalGraph (primingFile, dataFile, params, outFolder, logFiles)

Expand source code

class TemporalGraph(Graph):

    def __init__(self, primingFile, dataFile, params, outFolder, logFiles):
        
        params["relabel_nodes"] = False
        params["weighted"] = True
        
        Graph.__init__(self, primingFile, dataFile, params, outFolder, logFiles)
    
    def postProcess(self, G):

        nodes = set()
        for g in G:
            for n in list(g.nodes):
                nodes.add(n)
        nodes = list(nodes)

        mapping = {}
        i = 0
        for n in nodes:
            mapping[n]=i
            i += 1

        data = {
            "ASN":nodes,
            "edge_index":[],
            "edge_attr":[],
            "x":[]
        }

        for g in G:

            g_new = nx.relabel_nodes(g, mapping, copy=True)
            edges = list(g_new.edges)

            edge_index = torch.LongTensor(edges)
            edge_attr = torch.tensor([[g_new[u][v]["nbIp"]] for u,v in edges])

            x = []
            for i in range(len(nodes)):
                if(i in g_new.nodes):
                    x.append([g_new.nodes[i]["nbIp"]])
                else:
                    x.append([0])
            x = torch.tensor(x)

            data["edge_index"].append(edge_index)
            data["edge_attr"].append(edge_attr)
            data["x"].append(x)

        return(data)

Ancestors

• Graph
• BaseTransformParallelized
• BaseTransform
• BmlProcess

Methods

def postProcess(self, G)

Expand source code

def postProcess(self, G):

    nodes = set()
    for g in G:
        for n in list(g.nodes):
            nodes.add(n)
    nodes = list(nodes)

    mapping = {}
    i = 0
    for n in nodes:
        mapping[n]=i
        i += 1

    data = {
        "ASN":nodes,
        "edge_index":[],
        "edge_attr":[],
        "x":[]
    }

    for g in G:

        g_new = nx.relabel_nodes(g, mapping, copy=True)
        edges = list(g_new.edges)

        edge_index = torch.LongTensor(edges)
        edge_attr = torch.tensor([[g_new[u][v]["nbIp"]] for u,v in edges])

        x = []
        for i in range(len(nodes)):
            if(i in g_new.nodes):
                x.append([g_new.nodes[i]["nbIp"]])
            else:
                x.append([0])
        x = torch.tensor(x)

        data["edge_index"].append(edge_index)
        data["edge_attr"].append(edge_attr)
        data["x"].append(x)

    return(data)