private static WordsAndCounts RetrieveContextWordsForAllWordForms(string word, string langCode)
{
if (AllWordFormsContextCache.IsInCache(word, langCode))
{
WordsAndCounts cachedContextWordsAndCounts =
AllWordFormsContextCache.LoadFromCache(word, langCode);
return(cachedContextWordsAndCounts);
}
// Retieve and merge the contexts of all word forms of the given word
ICollection <string> allWordForms = LemmaDictionaryUtils.GetAllWordForms(word, langCode);
WordsAndCounts contextWordsAndCounts = new WordsAndCounts();
foreach (string wordForm in allWordForms)
{
WordsAndCounts wordFormContextWordsAndCounts =
RetrieveContextWords(wordForm, langCode);
contextWordsAndCounts.AddAll(wordFormContextWordsAndCounts);
}
// Normalize the counts
List <string> allContextWords = new List <string>();
allContextWords.AddRange(contextWordsAndCounts.Words);
foreach (string contextWord in allContextWords)
{
double count = contextWordsAndCounts[contextWord];
contextWordsAndCounts[contextWord] = count / allWordForms.Count;
}
AllWordFormsContextCache.AddToCache(word, langCode, contextWordsAndCounts);
return(contextWordsAndCounts);
}
private static double GetWordCountInContext(string word, String langCode, WordsAndCounts context)
{
if (useLemmatization)
{
double wordCount = 0;
var allWordFormsCollection =
LemmaDictionaryUtils.GetAllWordForms(word, langCode);
foreach (string wordForm in allWordFormsCollection)
{
wordCount += context[wordForm];
}
return(wordCount);
}
else
{
double wordCount = context[word];
return(wordCount);
}
}
private static Dictionary <string, bool> GetAllWordFormsDictionary(
string word, string langCode)
{
Dictionary <string, bool> allWordFormsDict = new Dictionary <string, bool>();
if (useLemmatization)
{
var allWordFormsCollection =
LemmaDictionaryUtils.GetAllWordForms(word, langCode);
foreach (string wordForm in allWordFormsCollection)
{
allWordFormsDict.Add(wordForm, true);
}
}
else
{
allWordFormsDict.Add(word, true);
}
return(allWordFormsDict);
}