// Due to lack of such basic feature as giving list of neighbors,
// this method builds two mappings:
// vertex label -> labels of broader vertices
// vertex label -> labels of narrower vertices
private void BuildNeighborsMappings()
{
// contains entries: v label -> set of labels of broaders
// Dictionary<string, HashSet<String>> brDict = new Dictionary<string, HashSet<String>>();
// // contains entries: v label -> set of labels of narrowers
// Dictionary<string, HashSet<String>> nrDict = new Dictionary<string, HashSet<String>>();
foreach (var arc in Graph.Arcs())
{
string src = "";
src = LabelsMap.TryGetValue(Graph.U(arc), out src) ? src : "";
src = src.ToLower();
string target = "";
target = LabelsMap.TryGetValue(Graph.V(arc), out target) ? target : "";
target = target.ToLower();
string broader = "";
string narrower = "";
string edgeType = "";
EdgeTypeMap.TryGetValue(arc, out edgeType);
if (edgeType == "broader")
{
// src - broader -> tgt
narrower = src;
broader = target;
}
else if (edgeType == "narrower")
{
// src - narrower -> tgt => tgt - broader -> src
broader = src;
narrower = target;
}
else
{
throw new Exception("Unknown relationship type found: '" + edgeType + "'");
}
if (IsAppropriateForIndex(broader) && IsAppropriateForIndex(narrower))
{
if (narrower != null && narrower.Trim() != "")
{
if (!Broaders.ContainsKey(narrower))
{
Broaders.Add(narrower, new HashSet <string>());
}
Broaders[narrower].Add(broader);
}
if (broader != null && broader.Trim() != "")
{
if (!Narrowers.ContainsKey(broader))
{
Narrowers.Add(broader, new HashSet <string>());
}
Narrowers[broader].Add(narrower);
}
}
}
Console.WriteLine("Loaded " + Broaders.Count + " broader relations");
Console.WriteLine("Loaded " + Narrowers.Count + " narrower relations");
}
// Build word -> labels index. It's crucial, because of big size of graph
// we cannot compare each word with all labels in graph
private void BuildWordIndex(int wordMinLen = 4)
{
foreach (var n in Graph.Nodes())
{
var label = LabelsMap[n];
// avoid adding labels which are not related with any document in index
if (label != null && IsAppropriateForIndex(label))
{
label = label.ToLower();
var words = Regex.Split(label, @"\s");
foreach (var w in words)
{
if (w.Length >= wordMinLen)
{
if (!WordLabels.ContainsKey(w))
{
WordLabels.Add(w, new HashSet <string>());
}
WordLabels[w].Add(label);
}
}
}
}
}