public void AddVertexToCommunity(EEVertex eev, EvalVal eval, double pScore)
{
if (!eval.memberable)
{
return;
}
if (!MemberNodes.ContainsKey(eev.Entity))
{
MemberNodes.Add(eev.Entity, eev.Neighbors.Count);
_pScoreSum += pScore;
}
this._outlinks += (uint)eval.linksOutCommunity;
this._volume += (uint)eev.Neighbors.Count;
if (OutlinkedNodes.ContainsKey(eev.Entity))
{
OutlinkedNodes.Remove(eev.Entity);
}
foreach (string nbor in eev.Neighbors.Keys)
{
string[] kws = eev.Neighbors[nbor].kw.Split(",".ToCharArray());
foreach (string kw in kws)
{
if (kw != "" && kw != " ")
{
if (!TrappedKeywords.ContainsKey(kw))
{
TrappedKeywords.Add(kw, 0);
}
TrappedKeywords[kw]++;
}
}
if (!isMember(nbor))
{
if (!OutlinkedNodes.ContainsKey(nbor))
{
OutlinkedNodes.Add(nbor, 1 /*eev.Neighbors[nbor].Weight*/);
}
}
}
}
public EvalVal EvalEEVertex(EEVertex eev)
{
EvalVal eval = new EvalVal();
eval.linksOutCommunity = 0;
eval.linksToCommunity = 0;
foreach (string nbor in eev.Neighbors.Keys)
{
if (MemberNodes.ContainsKey(nbor))
{
eval.linksToCommunity++;
}
else
{
eval.linksOutCommunity++;
}
}
eval.cond = ((double)this.OutlinksCount + eval.linksOutCommunity - eval.linksToCommunity) / (this.Volume + eev.Neighbors.Count);
if (eval.cond <= this.Conductance)
{
eval.memberable = true;
}
else if (eval.cond - CommCondThres < this.Conductance)
{
eval.memberable = true;
}
else
{
eval.memberable = false;
}
return(eval);
}
public void Communitize(string filepath, double comThres = 0.01, int MinNeighborForSeed = 2, string forcedSeed = "")
{
DateTime startTime = DateTime.Now;
Dictionary <string, int> AllEntity = new Dictionary <string, int>(StringComparer.InvariantCultureIgnoreCase);
foreach (string entity in EkwEgg.Keys)
{
if (EkwEgg[entity].Neighbors.Count >= MinNeighborForSeed)
{
if (!AllEntity.ContainsKey(entity))
{
AllEntity.Add(entity, EkwEgg[entity].Neighbors.Count);
}
}
}
bool keepRunning = true;
string seed = forcedSeed;
Console.WriteLine("building communities");
Console.WriteLine("SignificantEntity.Count: " + AllEntity.Count);
while (keepRunning)
{
if (AllEntity.Count < 1)
{
keepRunning = false; continue;
}
Community aCommunity = new Community();
aCommunity.CommCondThres = comThres;
//seed it
if (seed == "" || !AllEntity.ContainsKey(seed))
{
seed = AllEntity.Keys.First();
}
while (EkwEgg[seed].Neighbors.Count < MinNeighborForSeed)
{
AllEntity.Remove(seed);
seed = AllEntity.Keys.First();
Console.Write("\r" + AllEntity.Count);
}
Console.WriteLine("\t EntityRemain : " + AllEntity.Count);
bool keepBuildingThisComnty = true;
string entity = seed;
while (keepBuildingThisComnty)
{
EEVertex eev = EkwEgg[entity];
Community.EvalVal eval = aCommunity.EvalEEVertex(eev);
if (eval.memberable)
{
aCommunity.AddVertexToCommunity(eev, eval, EkwDict[entity].PScore);
AllEntity.Remove(entity);
Console.Write("\r" + AllEntity.Count);
}
else if (aCommunity.OutlinkedNodes.Count < 1)
{
keepBuildingThisComnty = false;
continue;
}
else
{
//this is not going to be a potential member
aCommunity.OutlinkedNodes.Remove(entity);
}
if (aCommunity.OutlinkedNodes.Count > 0)
{
entity = aCommunity.OutlinkedNodes.Keys.First();
}
else
{
keepBuildingThisComnty = false;
continue;
}
}
if (aCommunity.Size > 0)
{
aCommunity.Consolidate();
Communities.Add(aCommunity);
}
Console.Write("\tCommunity size:" + aCommunity.MemberNodes.Count);
seed = entity;
}
Console.WriteLine("time to find communitieis: " + (TimeSpan)(DateTime.Now - startTime));
int largesComntySz = 0;
StreamWriter fs = new StreamWriter(filepath);
fs.WriteLine("<Communities>");
int i = 1;
foreach (Community comnty in Communities)
{
//int kwThreshhold = comnty;
fs.WriteLine(" <Community id='" + i++ + "' size='" + comnty.Size + "' conductance='" + comnty.Conductance + "' pScore='" + comnty.pScore + "'>");
fs.WriteLine(" <trapped-keywords count='" + comnty.TrappedKeywords.Count + "'>");
fs.Write(" ");
foreach (string kw in comnty.TrappedKeywords.Keys)
{
fs.Write(System.Net.WebUtility.HtmlEncode(kw) + ":" + comnty.TrappedKeywords[kw] + ",");
}
fs.WriteLine("\n </trapped-keywords>");
foreach (string member in comnty.MemberNodes.Keys)
{
fs.WriteLine(" <e>" + System.Net.WebUtility.HtmlEncode(member) + "</e>");
}
fs.WriteLine(" </Community>");
largesComntySz = (largesComntySz > comnty.Size) ? largesComntySz : comnty.Size;
}
fs.WriteLine("</Communities>");
fs.Close();
Console.WriteLine("\n\nCommunities.Count: " + Communities.Count);
Console.WriteLine("Largest Communities Size: " + largesComntySz);
XmlDocument xDoc = new XmlDocument();
xDoc.Load(filepath);
xDoc.Normalize();
xDoc.Save(filepath);
}
