public FormMain()
{
InitializeComponent();
//_tc = TextClassificator.CreateLearnedInstance();
System.Threading.Thread.CurrentThread.CurrentCulture = new CultureInfo("ru-RU");
_tc = TextClassificator.LearnedInstance;
TextClassificator.GetLoadInstance("_ri");
TextClassificator.GetLoadInstance("_le");
TextClassificator.GetLoadInstance("_ei");
TextClassificator.GetLoadInstance("_si");
}
public static TextClassificator CreateLearnedInstance(string themeGroupName)
{
if (_learnedInstances == null)
_learnedInstances = new Dictionary<string, TextClassificator>();
if (!_learnedInstances.ContainsKey(themeGroupName))
{
var separators = new List<string>();
var fs = (string)Resource.ResourceManager.GetObject(TextClassificator.SeparatorsFileName);
separators = fs.Split(new[] { '\n' }, StringSplitOptions.RemoveEmptyEntries).ToList();
var stopWords = new List<string>();
fs = (string)Resource.ResourceManager.GetObject(TextClassificator.StopWordsFileName);
stopWords = fs.Split(new[] { '\n' }, StringSplitOptions.RemoveEmptyEntries).ToList();
var instance = new TextClassificator(separators, stopWords);
var themes = new Dictionary<string, string>();
instance.Learn(themeGroupName, themes);
_learnedInstances.Add(themeGroupName, instance);
}
return _learnedInstances[themeGroupName];
}
private static TextClassificator DefaultLoad(string groupPostfix)
{
var separators = new List<string>();
var fs = (string)Resource.ResourceManager.GetObject(TextClassificator.SeparatorsFileName);
separators = fs.Split(new[] { '\n' }, StringSplitOptions.RemoveEmptyEntries).ToList();
var stopWords = new List<string>();
fs = (string)Resource.ResourceManager.GetObject(TextClassificator.StopWordsFileName);
stopWords = fs.Split(new[] { '\n' }, StringSplitOptions.RemoveEmptyEntries).ToList();
var instance = new TextClassificator(separators, stopWords);
var files = new List<System.Collections.DictionaryEntry>();
var findString = groupPostfix;
foreach (System.Collections.DictionaryEntry file in Resource.ResourceManager.GetResourceSet(new System.Globalization.CultureInfo("ru-RU"), true, true))
if (file.Key.ToString().Length >= findString.Length && file.Key.ToString().Substring(file.Key.ToString().Length - findString.Length, findString.Length) == findString)
files.Add(file);
var filesPath = files.ToDictionary(f => f.Key.ToString().Substring(0, f.Key.ToString().IndexOf(groupPostfix)), f => Resource.ResourceManager.GetObject(f.Key.ToString()).ToString());
instance.ImportFromXmlsString(filesPath);
return instance;
}
/// <summary>создать обученный экземпляр классификатора (все значения берутся по умолчанию)
/// </summary>
/// <returns></returns>
public static TextClassificator CreateLearnedInstance()
{
var separators = new List<string>();
var fs = (string)Resource.ResourceManager.GetObject(TextClassificator.SeparatorsFileName);
separators = fs.Split(new[] { '\n' }, StringSplitOptions.RemoveEmptyEntries).ToList();
var stopWords = new List<string>();
fs = (string)Resource.ResourceManager.GetObject(TextClassificator.StopWordsFileName);
stopWords = fs.Split(new[] { '\n' }, StringSplitOptions.RemoveEmptyEntries).ToList();
var instance = new TextClassificator(separators, stopWords);
var themes = new Dictionary<string, string>();
foreach (System.Collections.DictionaryEntry file in Resource.ResourceManager.GetResourceSet(new System.Globalization.CultureInfo("ru-RU"), true, true))
{
if (file.Key.ToString()[0] == '_')
{
var text = file.Value.ToString();
text = text.ToLower();
text = DeleteWords(text, separators, false);
text = DeleteWords(text, stopWords);
text = DeleteDoubleSpaceAndBadSymbols(text);
themes.Add(file.Key.ToString().TrimStart('_'), text);
}
}
using (var db = new TextClassificatorEntities())
{
instance.Learn(db.ThemeGroup.Single(tg => tg.ThemeGroup_id == 1).ThemeGroup_name, themes);
}
return instance;
}
