/// <summary>
/// Gets the information the user wishes to analyse and prepares it for the network to analyse
/// </summary>
/// <param name="file">the file the user wishes to analyse</param>
public void GetAnalysedText(FileObj file)
{
FileReadWrite frw = new FileReadWrite();
CategoryObj analysingText = new CategoryObj(frw.GetStopWords(), frw.GetSuffixes());
analysingText.Name = file.FileName;
analysingText.AddText(file.FileContent);
_analysingText = analysingText;
}
/// <summary>
/// retrieves the networks from the folder.
/// </summary>
/// <returns>returns the bayesingNetwork array</returns>
public BayesingNetwork[] GetSavedBayesingNetworks()
{
//list of networks
List <BayesingNetwork> bayesingNetworks = new List <BayesingNetwork>();
try
{
//d is the directory which holds the networks infomation
foreach (string d in Directory.GetDirectories(_BayesingNetworkFolder))
{
//contains the categories for the networks
List <CategoryObj> cat = new List <CategoryObj>();
//file is the networks categories
foreach (string file in Directory.EnumerateFiles(d, "*.txt"))
{
CategoryObj c = new CategoryObj(GetStopWords(), GetSuffixes())
{
Name = Path.GetFileName(file)
};
//collects the dictionary information for the categories
string[] information = File.ReadAllLines(file);
c.DocumentsUsed = int.Parse(information[0]);
Dictionary <string, int> kvp = new Dictionary <string, int>();
for (int i = 1; i < information.Length; i++)
{
//splits the key from the value it holds
string[] WordAndCountSplit = information[i].Split('+');
int amountOfWords = int.Parse(WordAndCountSplit[1]);
kvp.Add(WordAndCountSplit[0], amountOfWords);
}
//new dictionary entry to be added to the category
c.WordInformation = kvp;
cat.Add(c);
}
BayesingNetwork bn = new BayesingNetwork(cat)
{
Name = d
};
bayesingNetworks.Add(bn);
}
}
catch (Exception)
{
Console.WriteLine("File error, please check the files are input correctly");
}
return(bayesingNetworks.ToArray());
}
/// <summary>
/// Trains the network
/// </summary>
public void Train()
{
FileReadWrite frw = new FileReadWrite();
//get the training files
FileObj[] files = frw.GetTrainingData();
List <CategoryObj> categories = new List <CategoryObj>();
bool exists = false;
//classifies the categories using the files
foreach (FileObj f in files)
{
string category = ""; //name of the new category
//creates the category name
foreach (char c in f.FileName)
{
if (c != '0' && c != '1' && c != '2' && c != '3' && c != '4' && c != '5' && c != '6' && c != '7' && c != '8' && c != '9' && c != '.')
{
category += c;
}
else
{
break;
}
}
exists = false;
//checks if information can be added to an existing category
foreach (CategoryObj cat in categories)
{
if (cat.Name == category)
{
exists = true;
cat.AddText(f.FileContent);
}
}
//makes a new category if the category doesn't already exist
if (!exists)
{
CategoryObj newCategory = new CategoryObj(frw.GetStopWords(), frw.GetSuffixes());
newCategory.Name = category;
newCategory.AddText(f.FileContent);
categories.Add(newCategory);
}
}
//adds the list of categorys tot he known information
_knownInformation = categories;
}