public override double ComputeWordProbability(string u, string v, string w)
{
Bigram uvBigram = new Bigram {
v = u, w = v
};
Trigram uvwTrigram = new Trigram {
u = u, v = v, w = w
};;
// Compute word probability given the previous two
this.NGramCounts.TryGetValue(uvBigram.GetComparisonKey(), out int uvCount);
this.NGramCounts.TryGetValue(uvwTrigram.GetComparisonKey(), out int uvwCount);
// q(w|u, v) = c(u, v, w)/c(u, v)
// q(w|u, v)_{addK} = (c(u, v, w) + k)/(c(u, v) + k|V*|)
double qWuv = this.Smoother.ComputeSmoothedWordProbability(u, v, w, uvwCount, uvCount, this.UniqueNGramsCount);
return(qWuv);
}
public override double ComputeWordProbability(string u, string v, string w)
{
Unigram vUnigram = new Unigram {
w = v
};
Bigram vwBigram = new Bigram {
v = v, w = w
};
// Compute word probability given the previous one
this.NGramCounts.TryGetValue(vUnigram.GetComparisonKey(), out int vCount);
this.NGramCounts.TryGetValue(vwBigram.GetComparisonKey(), out int vwCount);
// q(w|v) = c(v, w)/c(v)
// q(w|v)_{addK} = (c(v, w) + k)/(c(v) + k|V*|)
double qWv = this.Smoother.ComputeSmoothedWordProbability(u, v, w, vwCount, vCount, this.UniqueNGramsCount);
return(qWv);
}
public override void TrainLanguageModel(Corpus trainingCorpus)
{
// We need to process sentece by sentence to avoid wrapping sentences, ie. counting (STOP, v, w) trigrams
foreach (var sentence in trainingCorpus.AllTokenizedSentences)
{
// Initialize x_{-1}, x_{-2} to START
var u = "<s>";
var v = "<s>";
// We now need to store all counts of c(u, v, w) and c(v, w)
foreach (var w in sentence)
{
Bigram uvBigram = new Bigram {
v = u, w = v
};
Trigram uvwTrigram = new Trigram {
u = u, v = v, w = w
};
// +1 to current count, current will be 0 if not found, thus starting at 1 as expected
this.NGramCounts.TryGetValue(uvBigram.GetComparisonKey(), out int uvCount);
uvCount++;
this.NGramCounts[uvBigram.GetComparisonKey()] = uvCount;
var isNewNgram = !this.NGramCounts.TryGetValue(uvwTrigram.GetComparisonKey(), out int uvwCount);
uvwCount++;
this.NGramCounts[uvwTrigram.GetComparisonKey()] = uvwCount;
if (isNewNgram)
{
this.UniqueNGramsCount++;
}
// Replace previous tokens
u = v;
v = w;
}
}
}