public void StarterGramsFromMessage(string message, NGram starterGrams)
{
message = NormalizeInput(message);
string[] seperators = { " " };
string[] words = message.Split(seperators, StringSplitOptions.RemoveEmptyEntries);
bool prevWordIsEndOfSentence = false;
if (words.Length == 0 || words == null)
{
return;
}
for (int i = 0; i < words.Length; i++)
{
if (i == 0)
{
AddStarterGram(words[i], starterGrams);
}
if (IsEndOfSentence(words[i]))
{
prevWordIsEndOfSentence = true;
}
else if (prevWordIsEndOfSentence)
{
AddStarterGram(words[i], starterGrams);
}
}
}
private void AddStarterGram(string word, NGram starterGrams)
{
if (starterGrams.dictionary.ContainsKey(word))
{
int ponderation = (int)((starterGrams.Sum() / starterGrams.dictionary.Count) * 0.2);
if (ponderation < 1)
{
ponderation = 1;
}
starterGrams.dictionary[word].Frequency += ponderation;
starterGrams.Sort(false);
}
else
{
int ponderation = (int)((starterGrams.Sum() / starterGrams.dictionary.Count) * 1.5);
if (ponderation < 1)
{
ponderation = 1;
}
Sequence newKey = new Sequence();
newKey.Frequency = ponderation;
starterGrams.dictionary.Add(word, newKey);
starterGrams.Sort(true);
}
}
private void AddSuggestionsFromUniGram(NGram nGram)
{
int count = 0;
nGram.Sort();
foreach (KeyValuePair <string, Sequence> entry in nGram.orderedSequence)
{
//Should first take the most frequent x keys
//calcul de probabilit/ de base
if (results.Keys.Contains(entry.Key) == false)
{
results.Add(entry.Key, entry.Value.Frequency / nGram.Sum());
}
count += 1;
if (count >= 100)
{
break;
}
}
}
//string end
public void GramsFromMessage(int gramsize, string message, NGram gramCollection)
{
//clef, mot, frequence
//List<KeyValuePair<string,Sequence>> newGrams = new List<KeyValuePair<string, Sequence>>();
message = NormalizeInput(message);
string[] seperators = { " " };
string[] words = message.Split(seperators, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i <= words.Length - gramsize; i++)
{
string key = "";
string word = "";
if (gramsize - 1 > 0)
{ //nGrams
key = String.Join(" ", words.Skip(i).Take(gramsize - 1));
word = words[i + gramsize - 1];
if (gramCollection.dictionary.ContainsKey(key))
{
if (gramCollection.dictionary[key].dictionary.ContainsKey(word))
{
int ponderation = (int)((gramCollection.dictionary[key].Sum() / gramCollection.dictionary[key].dictionary.Count) * 0.2);
if (ponderation < 1)
{
ponderation = 1;
}
gramCollection.dictionary[key].dictionary[word].Frequency += ponderation;
gramCollection.dictionary[key].dictionary[word].Sort(false);
}
else
{
Sequence newSequence = new Sequence();
newSequence.Frequency = (int)((gramCollection.dictionary[key].Sum() / gramCollection.dictionary[key].dictionary.Count) * 1.5);
gramCollection.dictionary[key].dictionary.Add(word, newSequence);
gramCollection.dictionary[key].dictionary[word].Sort(true);
}
}
else
{
Sequence newKey = new Sequence();
Sequence newSequence = new Sequence();
newSequence.Frequency = 5;
newKey.dictionary = new Dictionary <string, Sequence>();
newKey.dictionary.Add(word, newSequence);
gramCollection.dictionary.Add(key, newKey);
gramCollection.Sort(true);
}
} //Unigrams
else
{
key = words[i];
if (gramCollection.dictionary.ContainsKey(key))
{
int ponderation = (int)((gramCollection.Sum() / gramCollection.dictionary.Count) * 0.2);
if (ponderation < 1)
{
ponderation = 1;
}
gramCollection.dictionary[key].Frequency += ponderation;
gramCollection.Sort();
}
else
{
int ponderation = (int)((gramCollection.Sum() / gramCollection.dictionary.Count) * 1.5);
if (ponderation < 1)
{
ponderation = 1;
}
Sequence newSequence = new Sequence();
newSequence.Frequency = ponderation;
gramCollection.dictionary.Add(key, newSequence);
gramCollection.Sort(true);
}
}
}
}
private void LoadFromFile(bool isPerso, string PREFIX, string POSTFIX)
{
string[] args = Environment.GetCommandLineArgs();
string folder = "";
var n = 1;
if (isPerso)
{
if (args != null && args.Length > 1 && args[1] != null)
{
folder = args[1];
//cherche des fichiers contenant un nombre >= 0
n = 0;
}
}
else
{
if (args != null && args.Length > 2 && args[2] != null)
{
folder = args[2];
//cherche des fichiers contenant un nombre >= 1
n = 1;
}
}
string line;
while (true)
{
var nGram = new NGram();
// Read the file load it line by line.
System.IO.StreamReader file;
var path = Environment.ExpandEnvironmentVariables(folder + PREFIX + n + POSTFIX);
try
{
file = new System.IO.StreamReader(path);
}
catch (FileNotFoundException e)
{
//Load les nGrams en ordre tant qu'il y en a sinon sort.
break;
}
while ((line = file.ReadLine()) != null)
{
var index = line.LastIndexOf(',');
var rawSequence = line.Substring(0, index);
var frequency = line.Substring(index + 1);;
String key;
var sequence = new Sequence();
if (n > 1)
{
var words = rawSequence.Split(' ');
var last = words.Last();
key = "";
for (var i = 0; i < words.Length - 1; ++i)
{
key = i > 0 ? key + " " + words[i] : words[i];
}
if (nGram.dictionary.Keys.Contains(key))
{
//ajout dans le subdictionary
sequence.Frequency = parseFrequency(frequency);
if (!nGram.dictionary[key].dictionary.Keys.Contains(last))
{
nGram.dictionary[key].dictionary.Add(last, sequence);
}
}
else
{
//creation du subdictionary et ajout de l'entree
sequence.dictionary = new Dictionary <string, Sequence>();
var temp = new Sequence();
temp.Frequency = parseFrequency(frequency);
if (!sequence.dictionary.Keys.Contains(last))
{
sequence.dictionary.Add(last, temp);
}
if (!nGram.dictionary.Keys.Contains(key))
{
nGram.dictionary.Add(key, sequence);
}
}
}
else
{
//Pour l'unigram et l'unigram de debut de phrase
key = rawSequence;
sequence.Frequency = parseFrequency(frequency);
if (!nGram.dictionary.Keys.Contains(key))
{
nGram.dictionary.Add(key, sequence);
}
}
}
file.Close();
nGram.Sort();
nGram.Sum();
if (n == 0)
{
this.nGramDebutPhrase = nGram;
}
else
{
if (isPerso)
{
this.nGramsPerso.Add(nGram);
}
else
{
this.nGramsPublic.Add(nGram);
}
}
n += 1;
}
}
