Exemplo n.º 1
0
 static void StemmAllText(ref List <string> AllWords)
 {
     for (int i = 0; i < AllWords.Count; i++)
     {
         List <string> words = AllWords[i].Split(' ').ToList <string>();
         AllWords[i] = "";
         for (int j = 0; j < words.Count; j++)
         {
             words[j]     = Stemming.TransformingWord(words[j]);
             AllWords[i] += words[j] + " ";
         }
     }
 }
Exemplo n.º 2
0
        //Составление списка слов по одному
        public static void BoolSearch(string request)
        {
            List <string> document = new List <string>();

            foreach (var docs in startDocs)
            {
                document.Add(docs);
            }
            StemmAllText(ref document);

            List <string> wordList = CreateWordList(ref document);

            DelStopWordsText(ref wordList);
            ListCreate(ref wordList);

            Dictionary <string, List <int> > invIndex = new Dictionary <string, List <int> >();

            foreach (var a in wordList)
            {
                //Console.WriteLine(a);
                invIndex.Add(a, new List <int>());
            }
            foreach (var a in invIndex)
            {
                for (int i = 0; i < document.Count; i++)
                {
                    if (document[i].Contains(a.Key))
                    {
                        a.Value.Add(i);
                    }
                }
            }
            Console.WriteLine("Инвертированный индекс:");
            foreach (var a in invIndex)
            {
                Console.Write(a.Key + "-");
                foreach (var b in a.Value)
                {
                    Console.Write(b + " ");
                }
                Console.WriteLine();
            }

            Console.WriteLine();

            Dictionary <string, List <int> > invIndexGood = new Dictionary <string, List <int> >();


            List <string> requestWords = request.ToLower().Split(' ').ToList <string>();

            for (int i = 0; i < requestWords.Count; i++)
            {
                requestWords[i] = Stemming.TransformingWord(requestWords[i]);
            }
            DelStopWordsText(ref requestWords);


            foreach (var requestWord in requestWords)
            {
                if (invIndex.ContainsKey(requestWord))
                {
                    foreach (var a in invIndex)
                    {
                        // Console.WriteLine(":"+a.Key+" "+requestWord+ a.Key.Length + " " + requestWord.Length);
                        if (a.Key == requestWord)
                        {
                            invIndexGood.Add(a.Key, a.Value);
                        }
                    }
                }
                else
                {
                    invIndexGood.Add(requestWord, new List <int>());
                }
            }

            Console.WriteLine("Булев поиск:");
            List <int> result = new List <int>();

            foreach (var goodInd in invIndexGood)
            {
                foreach (var number in goodInd.Value)
                {
                    int count = 0;
                    foreach (var goodInd2 in invIndexGood)
                    {
                        if (goodInd2.Value.Contains(number))
                        {
                            count++;
                        }
                    }
                    if (count == invIndexGood.Count)
                    {
                        if (!result.Contains(number))
                        {
                            result.Add(number);
                        }
                    }
                }
            }
            foreach (var a in result)
            {
                Console.WriteLine(startDocs[a]);
            }

            Console.WriteLine();
        }
Exemplo n.º 3
0
        static void CalcScore(string request)
        {
            List <string> document = new List <string>();

            foreach (var docs in startDocs)
            {
                document.Add(docs);
            }
            StemmAllText(ref document);



            List <double> scores       = new List <double>();
            List <string> requestWords = request.ToLower().Split(' ').ToList <string>();

            for (int i = 0; i < requestWords.Count; i++)
            {
                requestWords[i] = Stemming.TransformingWord(requestWords[i]);
            }

            DelStopWordsText(ref requestWords);

            //single


            for (int i = 0; i < document.Count; i++)
            {
                scores.Add(0);
                foreach (var a in requestWords)
                {
                    scores[i] += CountWords(a, document[i]);
                }
            }
            //pair

            for (int i = 0; i < document.Count; i++)
            {
                for (int j = 0; j < requestWords.Count - 1; j++)
                {
                    scores[i] += CountWords(requestWords[j] + " " + requestWords[j + 1], document[i]);
                }
            }
            //AllWords

            for (int i = 0; i < document.Count; i++)
            {
                bool f = true;
                foreach (var a in requestWords)
                {
                    if (!document[i].Contains(a))
                    {
                        f = false;
                    }
                }
                if (f)
                {
                    scores[i]++;
                }
            }

            string str = "";

            foreach (var a in requestWords)
            {
                str += a + " ";
            }
            for (int i = 0; i < document.Count; i++)
            {
                if (document[i].Contains(str))
                {
                    scores[i] += 1f / 350f;
                }
            }
            double        temp;
            string        tempStr;
            List <string> result = startDocs;

            for (int i = 0; i < scores.Count - 1; i++)
            {
                for (int j = i + 1; j < scores.Count; j++)
                {
                    if (scores[i] < scores[j])
                    {
                        tempStr   = result[i];
                        result[i] = result[j];
                        result[j] = tempStr;
                        temp      = scores[i];
                        scores[i] = scores[j];
                        scores[j] = temp;
                    }
                }
            }
            Console.WriteLine("Ранжирование с помощью Score:");
            for (int i = 0; i < scores.Count; i++)
            {
                Console.WriteLine(i + 1 + "." + result[i]);
            }
        }