private void displayResults(d.Cluster[] clusters, int topDocs, short[] toCluster)
{
d.ClusteringAnalyzer analyzer = new d.ClusteringAnalyzer(index, toCluster);
int docCount = 0;
for (int i = 0; i < clusters.Length; i++)
{
Console.WriteLine("CLUSTER " + (i + 1) + " documents=" +
clusters[i].DocumentCount + " intra-Similarity: " + analyzer.GetIntraDistance(clusters[i]));
IDictionaryEnumerator en = clusters[i].CommonTermIds.GetEnumerator();
while (en.MoveNext())
{
Console.Write(" " + index.GetTerm(Convert.ToInt32(en.Key)));
}
Console.WriteLine("\n");
en = clusters[i].DocIds.GetEnumerator();
int count = 0;
while (count < topDocs && en.MoveNext())
{
short docId = Convert.ToInt16(en.Key);
Console.WriteLine(" " + index.GetTitle(docId));
Console.WriteLine(" " + index.GetURL(docId));
Console.WriteLine();
count++;
}
docCount += clusters[i].DocumentCount;
}
Console.WriteLine("total clustered documents: " + docCount);
}
private void execCluster()
{
float w = Convert.ToSingle(tbxPagerank.Text);
d.Index index = new d.Index(getIndexDir());
d.VSSearcher searcher = new d.VSSearcher(index, w);
d.ResultDocument[] results = searcher.Search(tbxQuery.Text.Trim());
if (results != null && results.Length > 0)
{
int toCluster = Convert.ToInt32(tbxDocs.Text);
toCluster = Math.Min(toCluster, results.Length);
short[] docIds = new short[toCluster];
for (int i = 0; i < toCluster; i++)
{
docIds[i] = results[i].DocId;
}
int k = Convert.ToInt32(tbxKmeans.Text);
d.Cluster[] clusters;
if (radKmeans.Checked)
{
clusters = new d.KMeansClustering(index, k, true).GetClusters(docIds, 10);
}
else if (radBuckshot.Checked)
{
clusters = new d.BisectingClustering(index, k).GetClusters(docIds, 10);
}
else
{
clusters = new d.BisectingClustering(index, k).GetClusters(docIds, 10);
}
StringBuilder sb = new StringBuilder();
sb.AppendFormat("<p style='border-bottom:solid 1px #999999;'>Your search returned <b>{0}</b> results. " +
" Displaying the top {1} documents clustered into at most {2} clusters:</p>", results.Length, toCluster, k);
for (int i = 0; i < clusters.Length; i++)
{
sb.AppendFormat("<p style='margin-bottom:-10px;font-weight:bold;font-size:11pt;'>Cluster {0}", i + 1);
sb.Append("<p>Common terms: ");
IDictionaryEnumerator en = clusters[i].CommonTermIds.GetEnumerator();
while (en.MoveNext())
{
sb.AppendFormat("{0} ", index.GetTerm(Convert.ToInt32(en.Key)));
}
sb.Append("<p>");
en = clusters[i].DocIds.GetEnumerator();
int count = 0;
int topDocs = 3;
while (count < topDocs && en.MoveNext())
{
short docId = Convert.ToInt16(en.Key);
string url = "http://" + index.GetURL(docId).Replace("%%", "/");
sb.AppendFormat("<p style='margin-bottom:-10px;'><a style='font-size:11pt;' href='{0}'>{1}</a>", url, index.GetTitle(docId));
sb.AppendFormat("<p><a style='color:green;font-size:9pt' href='{0}'>{0}</a>", url);
count++;
}
sb.Append("<p style='border-bottom: solid 1px #999999;'>");
}
ltrResults.Text = sb.ToString();
}
else
{
ltrResults.Text = "There were no results";
}
}