public bool IsSpam(Corpus corpus)
{
var countAllOccurences = (double) CompleteCorpus.CountAllOccurences();
var beliefInCorpusBeingSpam = SpamCorpus.CountAllOccurences()/countAllOccurences;
corpus.ForEveryOccurenceOfEachWord(
word =>
{
var occurencesOfWord = SpamCorpus.CountOccurencesOf(word);
if(occurencesOfWord == 0)
{
beliefInCorpusBeingSpam = beliefInCorpusBeingSpam * .4;
}
else
{
var beliefInWordOccuringAtAll = CompleteCorpus.CountOccurencesOf(word)/countAllOccurences;
var beliefInEvidenceAndSpam = SpamCorpus.CountOccurencesOf(word)/countAllOccurences;
beliefInCorpusBeingSpam = beliefInCorpusBeingSpam * beliefInEvidenceAndSpam / beliefInWordOccuringAtAll;
}
});
Debug.WriteLine(String.Format("Belief: {0}", beliefInCorpusBeingSpam));
return beliefInCorpusBeingSpam > .16;
}
public bool Contains(Corpus corpus)
{
var result = true;
foreach (var word in corpus.Words)
result &= Words.ContainsKey(word.Key);
return result;
}
public void Add(Corpus corpus)
{
foreach (var pair in corpus.Words)
{
var word = pair.Key;
var count = pair.Value;
if (Words.ContainsKey(word))
Words[word]++;
else
Words.Add(word, count);
}
}
public double BeliefOf(string word, Corpus corpus)
{
var occurencesOfWordInCorpus = corpus.CountOccurencesOf(word);
var allOccurencesOfAllWords = corpus.CountAllOccurences();
return occurencesOfWordInCorpus / (double)allOccurencesOfAllWords;
}
public void SpamFound(Corpus corpus)
{
SpamCorpus.Add(corpus);
CompleteCorpus.Add(corpus);
}
