private void run(d.Clustering clustering, int toClusterCount, int topDocs)
{
d.VSSearcher vs = new d.VSSearcher(index, 0.6f);
while (true)
{
Console.WriteLine("Query: ");
string query = Console.ReadLine();
Console.WriteLine(query);
if (query == "")
{
break;
}
d.ResultDocument[] searchResults = vs.Search(query);
if (searchResults != null && searchResults.Length > 0)
{
int docs = Math.Min(searchResults.Length, toClusterCount);
short[] toCluster = new short[docs];
for (int i = 0; i < docs; i++)
{
toCluster[i] = searchResults[i].DocId;
}
displayResults(clustering.GetClusters(toCluster, 10), topDocs, toCluster);
}
else
{
Console.WriteLine("Your search returned 0 results -> there's nothing to cluster");
}
}
}
private void execSearch()
{
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)
{
StringBuilder sb = new StringBuilder();
sb.AppendFormat("<p style='border-bottom:solid 1px #999999;'>Your search returned <b>{0}</b> results. Displaying the top {1}:</p>", results.Length, tbxDisplay.Text);
sb.Append("<table style='font-family:Verdana;font-size:10pt;' cellpadding='3' cellspacing='2' width='100%'>");
sb.Append("<tr bgcolor='#f1f1f1'><td>#</td><td>similarity</td><td>pagerank</td><td>document</td></tr>");
int max = Math.Min(results.Length, Convert.ToInt32(tbxDisplay.Text));
for (int i = 0; i < max; i++)
{
int docId = results[i].DocId;
string url = "http://" + index.GetURL(docId).Replace("%%", "/");
sb.AppendFormat("<tr><td>{4}</td><td>{0}</td><td>{1}</td><td><p style='margin-bottom:-10px;'><a style='font-size:11pt;' href='{2}'>{3}</a>",
results[i].Similarity, results[i].PageRank, url, index.GetTitle(docId), i + 1);
sb.AppendFormat("<p><a style='color:green;font-size:9pt' href='{0}'>{0}</a></td></tr>", url);
}
sb.Append("</table>");
ltrResults.Text = sb.ToString();
}
else
{
ltrResults.Text = "There were no results";
}
}
public void run()
{
d.VSSearcher vs = new d.VSSearcher(index, 0.6f);
while (true)
{
Console.WriteLine("Query: ");
string query = Console.ReadLine();
if (query == "")
{
break;
}
d.ResultDocument[] searchResults = vs.Search(query);
if (searchResults != null && searchResults.Length > 0)
{
Console.WriteLine(searchResults.Length + " results:");
int max = Math.Min(10, searchResults.Length);
for (int i = 0; i < max; i++)
{
Console.WriteLine("\t" + searchResults[i].DocId + " " + searchResults[i].Similarity + " " +
index.GetURL(searchResults[i].DocId));
}
}
else
{
Console.WriteLine("Your search returned 0 results");
}
}
}
private void execAuthorities()
{
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)
{
StringBuilder sb = new StringBuilder();
int rootSize = Convert.ToInt32(tbxRoot.Text);
int maxParents = Convert.ToInt32(tbxParents.Text);
int maxChildren = Convert.ToInt32(tbxChildren.Text);
d.AHPageLoader ahl = new d.AHPageLoader(index, results, rootSize, maxParents, maxChildren);
d.AHDocument[] authorities = ahl.Authorities;
int max = Math.Min(authorities.Length, Convert.ToInt32(tbxDisplay.Text));
sb.AppendFormat("<p style='border-bottom:solid 1px #999999;'>Your search returned <b>{0}</b> results. " +
"Displaying the top {1} authorities:</p>", results.Length, max);
sb.Append("<table style='font-family:Verdana;font-size:10pt;' cellpadding='3' cellspacing='2' width='100%'>");
sb.Append("<tr bgcolor='#f1f1f1'><td>authority score</td><td>hub score</td><td>document</td></tr>");
for (int i = 0; i < max; i++)
{
int docId = authorities[i].DocId;
string url = "http://" + index.GetURL(docId).Replace("%%", "/");
sb.AppendFormat("<tr><td>{0}</td><td>{1}</td><td><p style='margin-bottom:-10px;'><a style='font-size:11pt;' href='{2}'>{3}</a>",
authorities[i].AuthorityScore, authorities[i].HubScore, url, index.GetTitle(docId));
sb.AppendFormat("<p><a style='color:green;font-size:9pt' href='{0}'>{0}</a></td></tr>", url);
}
sb.Append("</table>");
ltrResults.Text = sb.ToString();
}
else
{
ltrResults.Text = "There were no results";
}
}
public void run()
{
d.VSSearcher vs = new d.VSSearcher(index, 0.6f);
while (true)
{
Console.WriteLine("Query: ");
string query = Console.ReadLine();
if (query == "")
{
break;
}
d.ResultDocument[] searchResults = vs.Search(query);
if (searchResults != null && searchResults.Length > 0)
{
d.AHPageLoader ahl = new d.AHPageLoader(index, searchResults, 50, 20, 20); //if root>100, then bad results
d.AHDocument[] authorities = ahl.Authorities;
for (int i = 0; i < 20; i++)
{
Console.WriteLine(" a=" + authorities[i].AuthorityScore + " h=" + authorities[i].HubScore + " " + index.GetURL(authorities[i].DocId));
}
Console.WriteLine();
d.AHDocument[] hubs = ahl.Hubs;
for (int i = 0; i < 20; i++)
{
Console.WriteLine(" a=" + hubs[i].AuthorityScore + " h=" + hubs[i].HubScore + " " + index.GetURL(hubs[i].DocId));
}
}
else
{
Console.WriteLine("Your search returned 0 results");
}
}
}
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";
}
}
public void SetUp()
{
vs = new d.VSSearcher(new d.Index(EnvironmentBuilder.INDEX_DIRECTORY), 0.6f);
}