Пример #1
0
        public double[] transform(string sentence, string type)
        {
            sentence = RemoveSpecialCharacters(sentence);//.Replace("*", "").Replace(",", "").Replace("(", "").Replace(")", "");
            sentence = String.Join(" ", sentence.Split(' ').Where(x => !StopWords.IsMatch(x.ToLower().Trim())));

            if (type == "3-grams")
            {
                return(GetNGramsVector(sentence));
            }
            else if (type == "words")
            {
                return(GetWordsVector(sentence));
            }
            else
            {
                return(GetLettersVector(sentence));
            }
        }
Пример #2
0
        public void addSentence(string sentence)
        {
            sentence = RemoveSpecialCharacters(sentence);
            var sentenceWithoutStopWords = String.Join(" ", sentence.Split(' ').Where(x => !StopWords.IsMatch(x.ToLower().Trim())));

            string[] result = sentenceWithoutStopWords.Trim().Split(' ');
            // string sentenceWithoutStopWords = "";

            foreach (var r in result)
            {
                if (r.Length > 1)
                {
                    words.Add(r);
                }
                //sentenceWithoutStopWords += r;
            }
            createNGrams(sentenceWithoutStopWords.Replace(" ", ""));
        }