public industryLemmaRankTable GetTable()
{
industryLemmaRankTable output = null;
return(output);
}
public industryLemmaRankTable process(webLemmaTermTable chunkTable, webLemmaTermTable termTable, industryLemmaRankTable output)
{
List <webLemmaTerm> allChunks = chunkTable.GetList();
var docSetFreq = allChunks.Where(x => x.documentSetFrequency > 1);
instanceCountCollection <String> termCounter = new instanceCountCollection <string>();
aceDictionarySet <String, String> dict = new aceDictionarySet <string, string>();
foreach (webLemmaTerm chunk in docSetFreq)
{
var lemmas = chunk.nominalForm.SplitSmart(textMapBase.SEPARATOR, "", true, true);
lemmas = lemmas.Where(x => x.Length > 2).ToList();
termCounter.AddInstanceRange(lemmas);
foreach (String lm in lemmas)
{
foreach (String lmi in lemmas)
{
if (lmi != lm)
{
dict[lm].AddUnique(lmi);
}
}
}
}
List <String> primaries = new List <string>();
foreach (var pair in termCounter)
{
if (termCounter[pair] > 1)
{
primaries.Add(pair);
industryLemmaTerm lemma = output.GetOrCreate(pair);
lemma.termType = industryLemmaTermType.primary;
lemma.weight = settings.PrimaryTermFactor * termTable[lemma.name].weight;
lemma.nominalForm = pair;
output.AddOrUpdate(lemma);
if (dict.ContainsKey(lemma.nominalForm))
{
foreach (String secLemmas in dict[lemma.nominalForm])
{
industryLemmaTerm lemmaSec = output.GetOrCreate(secLemmas);
if (lemmaSec.termType == industryLemmaTermType.none)
{
lemmaSec.termType = industryLemmaTermType.secondary;
lemmaSec.weight = settings.SecondaryTermFactor * termTable[lemmaSec.name].weight;
lemmaSec.nominalForm = secLemmas;
output.AddOrUpdate(lemmaSec);
}
}
}
}
}
//var reserveChunks = allChunks.Where(x => x.nominalForm.ContainsAny(primaries));
//aceDictionarySet<String, String> dictReserve = new aceDictionarySet<string, string>();
//foreach (webLemmaTerm chunk in reserveChunks)
//{
// var lemmas = chunk.nominalForm.SplitSmart(textMapBase.SEPARATOR, "", true, true);
// lemmas = lemmas.Where(x => x.Length > 2).ToList();
// String prim = lemmas.FirstOrDefault(x => primaries.Contains(x));
// if (!prim.isNullOrEmpty())
// {
// foreach (String lm in lemmas)
// {
// if (prim != lm)
// {
// dictReserve[prim].AddUnique(lm);
// }
// }
// }
//}
//foreach (String prim in primaries)
//{
// if (dictReserve.ContainsKey(prim))
// {
// foreach (String res in dictReserve[prim])
// {
// industryLemmaTerm resLemma = output.GetOrCreate(res);
// if (resLemma.termType == industryLemmaTermType.none)
// {
// resLemma.nominalForm = res;
// resLemma.weight = settings.ReserveTermFactor *termTable[resLemma.name].weight;
// resLemma.termType = industryLemmaTermType.reserve;
// }
// output.AddOrUpdate(resLemma);
// }
// }
//}
return(output);
}