// Extract and store features
public static void extractFeatures(string folderPath, int M, string output = "articles.feat")
{
DocCollection collection = new DocCollection();
// Collect all files
foreach (string filepath in Directory.GetFiles(folderPath, "*.txt"))
{
string content = File.ReadAllText(filepath);
collection.collect(content);
}
List<DocVector> docVect = collectionProcessing(collection, 20);
// Store features
using (StreamWriter writer = new StreamWriter(output, false))
{
writer.WriteLine(Directory.GetFiles(folderPath, "*.txt").Count()); // Number of files
writer.WriteLine(M); // Number of keywords
// Store bag of words
for (int i = 0; i < M; ++i)
writer.Write(keywords[i] + " ");
writer.WriteLine();
// Store feature for each document
foreach (DocVector vector in docVect)
{
for (int i = 0; i < M; ++i)
writer.Write(vector.Tf_idf[i] + " ");
writer.WriteLine();
}
}
}
