public static DataTable GetClassKnowledgeReport(this DocumentSetPipelineCollection pipelineCollection, IDocumentSetClass caseSet, DataTable output = null)
{
if (output == null)
{
output = new DataTable();
output.SetTitle(caseSet.name);
output.Add("Name", "Name of class or web site", "", typeof(String), imbSCI.Core.enums.dataPointImportance.normal, "", "Name").SetGroup("Repository").SetWidth(25).SetUnit("");
output.Add("Sites", "Number of sites in cateogry - or 1 for single site entry", "|C_ds|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "", "Sites").SetGroup("Repository").SetWidth(10).SetUnit("n");
output.Add("Pages", "Total number of pages detected in the repository", "|C_d|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "", "Pages Crawled").SetGroup("Repository").SetWidth(15).SetUnit("n");
output.Add("PagesValid", "Number of pages used for the category or site", "|C_dv|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "", "Pages Used").SetGroup("Pipeline").SetWidth(15).SetUnit("n");
output.Add("Blocks", "Number of blocks for category or site", "|C_b|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "").SetGroup("Pipeline").SetWidth(10).SetUnit("n");
output.Add("Streams", "Number of streams for category or site", "|C_ts|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "").SetGroup("Pipeline").SetWidth(10).SetUnit("n");
output.Add("Tokens", "Number of tokens for category or site", "|C_t|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "").SetGroup("Pipeline").SetWidth(10).SetUnit("n");
output.Add("Chunks", "Number of chunks for category - disabled for sites", "|C_c|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "").SetGroup("NLP").SetWidth(10).SetUnit("n");
output.Add("OnlyLetters", "Number of tokens for category or site with onlyLetter tag", "|C_ttl|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "", "Only Letters").SetGroup("Only Letters").SetWidth(10).SetUnit("n");
output.Add("OnlyLettersResolved", "Number of tokens resolved by morphosyntactic resource", "|C_ttlr|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "", "Accepted Tokens").SetGroup("Only Letters").SetWidth(10).SetUnit("n");
output.Add("OnlyLettersUnresolved", "Number of tokens unresolved by morphosyntactic resource", "|C_ttlu|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "", "Dismissed").SetGroup("Only Letters").SetWidth(10).SetUnit("n");
output.Add("Numbers", "Number of tokens for category or site with a numeric content tag", "|C_ttn|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "", "Number").SetGroup("Other").SetWidth(10).SetUnit("n");
output.Add("Symbols", "Number of tokens for category or site with a symbolic content tag", "|C_tts|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "", "Symbols").SetGroup("Other").SetWidth(10).SetUnit("n");
output.Add("Business", "Number of tokens for category or site with any dat_business tag", "|C_ttb|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "", "Business tags").SetGroup("Special").SetWidth(10).SetUnit("n");
output.Add("Potential", "Number of tokens for category or site with any tkn_potential data point tag", "|C_ttp|", typeof(Int32), imbSCI.Core.enums.dataPointImportance.normal, "", "Potential tags").SetGroup("Special").SetWidth(10).SetUnit("n");
}
else
{
output.SetTitle("Class Set Report");
output.TableName = "multi_class_report";
}
output.SetAdditionalInfoEntry("Class " + caseSet.treeLetterAcronim + " name", caseSet.name);
output.SetAdditionalInfoEntry("Class " + caseSet.treeLetterAcronim + " repo", caseSet.MCRepositoryName);
var sites = pipelineCollection.sitesByCategory[caseSet].ToList();
PipelineReportForClass repForClass = new PipelineReportForClass();
repForClass.deployClass(caseSet);
repForClass.Chunks = pipelineCollection.chunksByCategory[caseSet].Count();
foreach (var site in sites)
{
var dr = output.NewRow();
PipelineReportForClass repForSite = new PipelineReportForClass();
repForSite.deploySite(site);
repForSite.deployTokens(pipelineCollection.tokenBySite[site as pipelineTaskMCSiteSubject]);
repForSite.SetDataRow(dr);
output.Rows.Add(dr);
repForClass.sum(repForSite);
}
var drc = output.NewRow();
repForClass.SetDataRow(drc);
output.Rows.Add(drc);
output.GetRowMetaSet().AddUnit(new dataValueMatchCriterionDynamicStyle <String, DataRowInReportTypeEnum>(new String[] { repForClass.name }, DataRowInReportTypeEnum.dataHighlightA, "Name"));
return(output);
}
public void sum(PipelineReportForClass B)
{
Sites += B.Sites;
Pages += B.Pages;
PagesValid += B.PagesValid;
Blocks += B.Blocks;
Streams += B.Streams;
Chunks += B.Chunks;
Tokens += B.Tokens;
OnlyLetters += B.OnlyLetters;
OnlyLettersResolved += B.OnlyLettersResolved;
OnlyLettersUnresolved += B.OnlyLettersUnresolved;
Numbers += B.Numbers;
Symbols += B.Symbols;
Business += B.Business;
Potential += B.Potential;
}
