public void Invert()
{
var dict = new Dictionary <string, IList <int> >();
var index = fileProvider.GetTextFromFile($"{ParentForderPath}/{uri.Host}/index.txt");
var files = index.Split("\n").Select(x => x.Split(" ")).Select(x => x[2]).ToList();
for (int i = 0; i < files.Count; i++)
{
var words = fileProvider.GetTextFromFile($"{StemingForder}{i}.txt").Split(' ').ToList();
foreach (var word in words)
{
if (dict.ContainsKey(word))
{
dict[word].Add(i);
}
else
{
dict.Add(word, new List <int> {
i
});
}
}
}
foreach (var item in dict)
{
dict[item.Key] = item.Value.Distinct().ToList();
}
var invertDict = JsonConvert.SerializeObject(dict);
fileProvider.WriteTextToFile($"{StemingForder}invertlist.txt", invertDict);
}
public void TF()
{
for (int i = 0; i < docNum; i++)
{
var dict = new Dictionary <string, int>();
var text = fileProvider.GetTextFromFile($"{StemingForder}{i}.txt");
var words = fileProvider.GetTextFromFile($"{StemingForder}{i}.txt").Split(' ').ToList();
foreach (var word in words)
{
if (dict.ContainsKey(word))
{
dict[word]++;
}
else
{
dict.Add(word, 1);
}
}
var allWords = dict.Select(x => x.Value).Sum();
var result = string.Join('\n', dict.Select(x => $"{x.Key} : {Math.Round((double)x.Value / allWords,5).ToString("0.00000")}").ToList());
fileProvider.WriteTextToFile($"{tfForder}{i}.txt", result);
}
}
private void StemmingFile(string path)
{
var text = fileProvider.GetTextFromFile(path);
var words = new List <string>();
MatchCollection collection = Regex.Matches(text, @"([\w]{1,})");
var porter = new Porter();
foreach (Match word in collection)
{
string stremmed;
if (word.Value.Length > 4)
{
stremmed = porter.Stemm(word.Value);
}
else
{
stremmed = word.Value;
}
words.Add(stremmed);
}
var filename = Regex.Match(path, @"([\d]*.txt)");
fileProvider.WriteTextToFile($"{ParentForderPath}/{uri.Host}/stemming/{filename.Value}", string.Join(' ', words));
}