public virtual double GetPhraseWeightFromWords(ICounter <CandidatePhrase> weights, CandidatePhrase ph, double defaultWt)
{
string[] t = ph.GetPhrase().Split("\\s+");
if (t.Length < 2)
{
if (weights.ContainsKey(ph))
{
return(weights.GetCount(ph));
}
else
{
return(defaultWt);
}
}
double totalscore = 0;
double minScore = double.MaxValue;
foreach (string w in t)
{
double score = defaultWt;
if (weights.ContainsKey(CandidatePhrase.CreateOrGet(w)))
{
score = weights.GetCount(w);
}
if (score < minScore)
{
minScore = score;
}
totalscore += score;
}
if (useAvgInsteadofMinPhraseScoring)
{
return(totalscore / ph.GetPhrase().Length);
}
else
{
return(minScore);
}
}
public static double GetGoogleNgramScore(CandidatePhrase g)
{
double count = GoogleNGramsSQLBacked.GetCount(g.GetPhrase().ToLower()) + GoogleNGramsSQLBacked.GetCount(g.GetPhrase());
if (count != -1)
{
if (!Data.rawFreq.ContainsKey(g))
{
//returning 1 because usually lower this tf-idf score the better. if we don't have raw freq info, give it a bad score
return(1);
}
else
{
return((1 + Data.rawFreq.GetCount(g) * Math.Sqrt(Data.ratioGoogleNgramFreqWithDataFreq)) / count);
}
}
return(0);
}
