public CorpusAnalyser(IEnumerable<string> samples)
{
_index = new DocumentIndex();
_markovChain = new DiscreteMarkovChain<string>();
_documentTermMatrix = Analyse(samples.Select(s => new TokenisedTextDocument(Guid.NewGuid().ToString(), _index.Tokeniser.Tokenise(s))));
}
public CorpusAnalyser(IEnumerable<TokenisedTextDocument> samples)
{
_index = new DocumentIndex();
_markovChain = new DiscreteMarkovChain<string>();
_documentTermMatrix = Analyse(samples);
}
public void SecondOrder_AddSequence_and_GetFrequencies_ReturnsCorrectFrequency(string sequence, char test, char assertChar, int assertFreq, int totalCount)
{
var mkc = new DiscreteMarkovChain<char>(2);
mkc.AnalyseSequence(sequence);
var freq = mkc.GetFrequencies(test);
Assert.That(freq.Count, Is.EqualTo(totalCount));
Assert.That(freq[assertChar], Is.EqualTo(assertFreq));
}
public void AddSequence_and_Prune()
{
var mkc = new DiscreteMarkovChain<char>();
mkc.AnalyseSequence("a1b1c1b2c2b3");
Assert.That(mkc.GetFrequencies('b').Count, Is.EqualTo(3));
Assert.That(mkc.GetFrequencies('a').Count, Is.EqualTo(1));
mkc.Prune(3);
Assert.That(mkc.GetFrequencies('b').Count, Is.EqualTo(3));
Assert.That(mkc.GetFrequencies('a').Count, Is.EqualTo(0));
}
public void AddSequence_and_Merge_WithOther()
{
var mkc = new DiscreteMarkovChain<char>();
var mkc2 = new DiscreteMarkovChain<char>();
mkc.AnalyseSequence("a1b1c1b2c2b3");
mkc2.AnalyseSequence("b5b1");
Assert.That(mkc.GetFrequencies('b').Count, Is.EqualTo(3));
Assert.That(mkc.GetFrequencies('b')['1'], Is.EqualTo(1));
mkc.Merge(mkc2);
Assert.That(mkc.GetFrequencies('b').Count, Is.EqualTo(4));
Assert.That(mkc.GetFrequencies('b')['1'], Is.EqualTo(2));
}
public void AddSequence_GetFrequencies_ReturnsExpectedValues()
{
var mkc = new DiscreteMarkovChain<char>(2);
mkc.AnalyseSequence("abcabdabcabcabn");
var freq = mkc.GetFrequencies("ab");
Assert.That(freq.Count, Is.EqualTo(3));
Assert.That(freq['c'], Is.EqualTo(3));
Assert.That(freq['d'], Is.EqualTo(1));
Assert.That(freq['n'], Is.EqualTo(1));
}
public void AddSequence_ProbabilityOf()
{
var mkc = new DiscreteMarkovChain<char>(2);
var seq = "abcabdabcabcabn";
mkc.AnalyseSequence(seq);
var p = mkc.ProbabilityOfEvent("ab", 'c');
Assert.That(p.Value, Is.EqualTo(3d / 5d));
}
public void AddSequence_Simulate()
{
var mkc = new DiscreteMarkovChain<char>(2);
var seq = "abcabdabcabcabn";
mkc.AnalyseSequence(seq);
var simulation = mkc.Simulate('b').ToList();
Assert.That(simulation.Count > 0);
Assert.That(simulation.All(c => seq.IndexOf(c) > -1));
}
