private bool BelongsTo(Document document, DocumentCluster cluster)
{
foreach (Document doc in cluster.Documents)
{
double oddsRatio = _similarity.CalculateOddsRatio(document, doc);
if (oddsRatio >= 2)
{
return(true);
}
else
{
return(false);
}
}
return(false);
}
public IEnumerable <DocumentCluster> Cluster(IEnumerable <Document> documents)
{
List <DocumentCluster> clusters = new List <DocumentCluster>();
DocumentCluster cluster = new DocumentCluster();
foreach (Document document in documents)
{
cluster = FindCluster(document, clusters);
if (null == cluster)
{
cluster = new DocumentCluster();
clusters.Add(cluster);
}
cluster.Add(document);
}
return(clusters);
}
private static IEnumerable <DocumentClusterErrorScore> CalculateErrorScore(IEnumerable <DocumentCluster> resultClusters, IEnumerable <DocumentCluster> originalClusters)
{
List <DocumentClusterErrorScore> result = new List <DocumentClusterErrorScore>();
List <DocumentCluster> resultClustersList = resultClusters.ToList();
List <DocumentCluster> originalClustersList = originalClusters.ToList();
for (int i = 0; i < originalClustersList.Count; i++)
{
DocumentCluster originalCluster = originalClustersList[i];
int? resultClusterIndex = null;
foreach (Document originalDocument in originalCluster)
{
DocumentClusterErrorScore score = new DocumentClusterErrorScore();
int index = FindClusterIndex(resultClustersList, originalDocument);
if (index == -1)
{
throw new ArgumentException("Can't find doc");
}
else if (!resultClusterIndex.HasValue)
{
resultClusterIndex = index;
}
else if (index != resultClusterIndex.Value)
{
score.Value = -1;
}
else
{
score.Value = 1;
}
result.Add(score);
}
}
return(result.AsEnumerable());
}
