/// <summary>
/// Merge two lists of words together summing the frequency on same keys
/// </summary>
/// <param name="words1">First words list</param>
/// <param name="words2">Second words list</param>
/// <returns></returns>
private List <Word> MergeWords(List <Word> words1, List <Word> words2)
{
if (words1 == null)
{
return(words2);
}
else if (words2 == null)
{
return(words1);
}
List <Word> concatWords = words1.Concat(words2).ToList();
List <Word> returnWords = new List <Word>();
List <IGrouping <string, Word> > groups = concatWords.GroupBy(x => x.Key).ToList();
foreach (IGrouping <string, Word> group in groups)
{
Word first = group.FirstOrDefault();
first.SetFrequency(group.Sum(x => x.Frequency));
first.DocumentFrequency = group.Sum(x => x.DocumentFrequency);
returnWords.Add(first);
}
return(returnWords);
}
/// <summary>
/// Get frequencies for all words in word list
/// </summary>
/// <param name="words">List of words to get frequencies for</param>
/// <returns></returns>
public List <Word> GetWordFrequencies(List <Word> words)
{
Dictionary <string, int> wordFrequencies = new Dictionary <string, int>();
List <Word> wordList = words.GroupBy(x => x.Key).Select(w => w.First()).ToList();
foreach (Word word in words)
{
if (wordFrequencies.TryGetValue(word.Key, out int freq))
{
wordFrequencies[word.Key] = freq + 1;
}
else
{
wordFrequencies[word.Key] = 1;
}
}
foreach (KeyValuePair <string, int> kvp in wordFrequencies)
{
Word word = wordList.Where(x => x.Key == kvp.Key).First();
word.SetFrequency(kvp.Value);
}
return(wordList);
}
