public double CalculateCosineSimilarity(CosineSimilarity cosineSimilarity, string documentA, string documentB, string documentAIndex, string documentBIndex)
{
var document1 = cosineSimilarity.MatrixMovie.Where(m => m.Document.ToLower() == documentA.ToLower());
var document2 = cosineSimilarity.MatrixMovie.Where(m => m.Document.ToLower() == documentB.ToLower());
double item1 = 0;
double sqrtItem1 = 0;
double sqrtItem2 = 0;
for (var i = 0; i < cosineSimilarity.BagOfWords.Count; i++)
{
item1 += document1.ElementAt(i).Exist *document2.ElementAt(i).Exist;
sqrtItem1 += Math.Pow(document1.ElementAt(i).Exist, 2);
sqrtItem2 += Math.Pow(document2.ElementAt(i).Exist, 2);
}
var finalResult = item1 / (Math.Sqrt(sqrtItem1) * Math.Sqrt(sqrtItem2));
foreach (var item in cosineSimilarity.BagOfWords)
{
Console.Write("|" + item);
}
Console.WriteLine();
Console.Write(string.Concat("DOCUMENTO BINÁRIO ", documentAIndex, ": "));
foreach (var item in document1)
{
Console.Write(" | " + item.Exist);
}
Console.WriteLine();
Console.Write(string.Concat("DOCUMENTO BINÁRIO ", documentBIndex, ": "));
foreach (var item in document2)
{
Console.Write(" | " + item.Exist);
}
Console.WriteLine();
Console.WriteLine(string.Concat("SIM(", documentAIndex, ",", documentBIndex, ") = ", finalResult));
Console.WriteLine();
return(finalResult);
}
static void Main(string[] args)
{
var movies = new Movies();
var stopWords = new StopWords();
var bagOfWordsCalc = new BagOfWordsCalc();
var moviesItems = movies.Get().ToList();
var stopWordsItens = stopWords.Get().ToList();
var finalWords = new List <string>();
foreach (var item in moviesItems)
{
foreach (var caracther in stopWords.CharactersToRemove())
{
item.Movie = item.Movie.Replace(caracther, "");
}
item.MovieSplited.AddRange(item.Movie.Split(' ').ToList());
}
foreach (var item in moviesItems)
{
var moviesWithoutStopWords = item.MovieSplited.Where(a => !stopWordsItens.Contains(a)).ToList();
item.MovieSplited = moviesWithoutStopWords;
finalWords.AddRange(item.MovieSplited.Where(a => !finalWords.Contains(a)));
}
var matrixMovies = new List <MatrixMovie>();
foreach (var item in moviesItems)
{
foreach (var finalWord in finalWords)
{
var exists = item.MovieSplited.Contains(finalWord)? 1: 0;
matrixMovies.Add(new MatrixMovie {
Document = item.Movie, Word = finalWord, Exist = exists
});
}
}
var cosineSimilarity = new CosineSimilarity {
MatrixMovie = matrixMovies, BagOfWords = finalWords
};
var documents = moviesItems.Select(a => a.Movie).ToList();
var result = Combinations.GetPermutations(documents, 2).ToList();
foreach (var item in result)
{
var itens = new List <string>();
foreach (var item2 in item)
{
itens.Add(item2);
}
var document1Index = moviesItems.FirstOrDefault(a => a.Movie == itens[0])?.Index;
var document2Index = moviesItems.FirstOrDefault(a => a.Movie == itens[1])?.Index;
var bagCalculation = bagOfWordsCalc.CalculateCosineSimilarity(cosineSimilarity, itens[0], itens[1], document1Index, document2Index);
}
Console.WriteLine();
Console.WriteLine("TABELA FREQUENCIA");
Console.WriteLine();
foreach (var word in finalWords)
{
Console.Write("|" + word);
}
Console.WriteLine();
foreach (var item in moviesItems)
{
Console.Write(item.Movie + ": ");
foreach (var word in finalWords)
{
Console.Write(" | " + item.MovieSplited.Count(f => word.Contains(f)));
}
Console.WriteLine();
}
Console.WriteLine();
Console.ReadLine();
}
