public void RefineStyle(classificationReportStyleDefinition style, IEnumerable <classificationReportExpanded> reports)
{
local_style = style.CloneViaXML();
if (local_style.AutoNewDataColumns || local_style.AutoHideDataColumnsWithSameData)
{
aceDictionarySet <String, String> knownValues = new aceDictionarySet <string, string>();
var expData = reports.Select(x => x.data);
foreach (var cd in expData)
{
foreach (var cl in cd)
{
if (local_style.AutoNewDataColumns)
{
if (!local_style.dataColumns.Any(x => x.key == cl.key))
{
local_style.dataColumns.Add(cl);
}
}
if (!knownValues[cl.key].Contains(cl.value))
{
knownValues.Add(cl.key, cl.value);
}
}
}
if (local_style.AutoHideDataColumnsWithSameData)
{
List <String> columnsToHide = new List <string>();
foreach (var pair in knownValues)
{
if (pair.Value.Count < 2)
{
columnsToHide.Add(pair.Key);
}
else if (pair.Value.Count == 0)
{
columnsToHide.Add(pair.Key);
}
}
var toRemove = local_style.dataColumns.Where(x => columnsToHide.Contains(x.key)).ToList();
toRemove.ForEach(x => local_style.dataColumns.Remove(x));
}
}
}
public static classificationReportStyleDefinition GetDefault(ExperimentRunNameGroups _groups = null)
{
classificationReportStyleDefinition output = new classificationReportStyleDefinition();
if (_groups == null)
{
_groups = new ExperimentRunNameGroups();
_groups.CheckForDefault();
}
output.layerNeedleByName.Add("Trashold", @"([\d]+)#([\d]*)", "Experiments with document selection controled by trashold");
output.groups = _groups;
output.dataFlags.Add("Render", "LinkText,LinkContent,LINCAPT", "A", "Content from link anchor text");
output.dataFlags.Add("Render", "Tokens,TKN", "TKN", "Tokens extracted from URL");
output.dataFlags.Add("Render", "PageText,PageContent", "B", "Page body text, page description and title tag");
output.dataFlags.Add("Scope", "category,InCategory,Category", "C", "Items in the category");
output.dataFlags.Add("Scope", "page", "P", "Pages");
output.dataFlags.Add("Scope", "Dataset,InDataset", "D", "Items in the complete dataset");
output.dataFlags.Add("Scope", "Link", "L", "Links");
output.dataFlags.Add("Function", "selfCentric", "SC", "Compares items with web site / document set");
output.dataFlags.Add("Function", "Inverse", "I", "Score value is inversed at the end of computation");
output.dataFlags.Add("Function", "ENT", "*", "Score value is inversed at the end of computation");
output.dataFlags.Add("Function", "LNG", "*", "Score value is inversed at the end of computation");
output.dataFlags.Add("Function", "DST", "*", "Score value is inversed at the end of computation");
output.dataFlags.Add("Function", "Divergence", "DIV", "Promotes diversity, items are at greater distance in vector space");
output.dataFlags.Add("Function", "Convergence", "CON", "Promotes convergence of the items, ones at smaller distance are promoted");
output.dataFlags.Add("Function", "Variance", "VAR", "Promotes variance of the items");
output.dataFlags.Add("Function", "Offset", "OFF", "Measures difference between similarity with true label and average similariy with other labels");
//output.dataColumns.Add("Render", "", "Source of the rendered");
//output.dataColumns.Add("Scope", "", "Scope of analysis, performed by the function");
//output.dataColumns.Add("Function", "", "Function performing the analysus");
//output.dataColumns.Add("WeightModel", "", "Model used for term weighting");
return(output);
}
/// <summary>
/// Builds the report space from report collection
/// </summary>
/// <param name="reports">The reports.</param>
/// <param name="dataset">The dataset.</param>
/// <param name="SELECT_REPORT_NAME_PARTS">The select report name parts.</param>
/// <returns></returns>
public static classificationReportSpace BuildReportSpace(IEnumerable <classificationReportExpanded> reports, String dataset, Regex SELECT_REPORT_NAME_PARTS, classificationReportStyleDefinition style, String _layer = "")
{
classificationReportSpace space = new classificationReportSpace()
{
dataset = dataset,
name = _layer.or(classificationReportExpanded.LAYER_MAIN)
};
foreach (var rep in reports)
{
String nm = rep.Name;
if (nm.StartsWith(dataset))
{
rep.data.Add("Dataset", dataset, "Confirmed dataset name");
nm = nm.Substring(dataset.Length);
}
Match mc = SELECT_REPORT_NAME_PARTS.Match(nm);
if (mc.Groups.Count > 1)
{
String runName = mc.Groups[1].Value.Trim('_');
Int32 size = 0;
if (mc.Groups.Count > 2)
{
String sizeString = mc.Groups[2].Value;
if (sizeString.isNullOrEmpty())
{
size = 0;
}
else
{
size = Int32.Parse(sizeString);
}
}
runName = style.dataFlags.ProcessRunName(runName, rep.data);
String nRunName = space.SetF1(runName, size, rep);
}
}
space.RefineStyle(style, reports);
space.Complete();
return(space);
}
public static Dictionary <String, List <classificationReportExpanded> > GetSpaceLayers(this IEnumerable <classificationReportExpanded> reports, classificationReportStyleDefinition style, Dictionary <String, List <classificationReportExpanded> > output = null)
{
if (output == null)
{
output = new Dictionary <string, List <classificationReportExpanded> >();
}
style.Prepare();
foreach (classificationReportExpanded rep in reports)
{
foreach (var pair in style.layerNeedleByNameCompiled)
{
if (pair.Key.IsMatch(rep.Name))
{
Match mc = pair.Key.Match(rep.Name);
String nm = rep.Name;
var splits = pair.Key.Split(nm);
nm = "";
foreach (String sp in splits)
{
nm += sp;
}
rep.layer = pair.Value;
}
}
if (!output.ContainsKey(rep.layer))
{
output.Add(rep.layer, new List <classificationReportExpanded>());
}
output[rep.layer].Add(rep);
}
return(output);
}
