public void ExtendDatabaseWithArticle(LabeledArticle article)
{
Annotation annotation = _processor.Annotate(article.Article);
HashSet <string> persons = NLPCoreHelper.GetPersons(annotation);
HashSet <string> organizations = NLPCoreHelper.GetOrganizations(annotation);
HashSet <string> locations = NLPCoreHelper.GetLocation(annotation);
HashSet <string> sportSpecificWords = NLPCoreHelper.GetOccurence(annotation, _sportSpecificWords);
foreach (string person in persons)
{
PersonOccurenceDatabase.AddWord(person, article.Category);
}
foreach (string organization in organizations)
{
OrganizationOccurenceDatabase.AddWord(organization, article.Category);
}
foreach (string location in locations)
{
LocationsOccurenceDatabase.AddWord(location, article.Category);
}
foreach (string sportSpecificWord in sportSpecificWords)
{
SportSpecificWordsOccurenceDatabase.AddWord(sportSpecificWord, article.Category);
}
}
private double[] GetRawFeatures(string article)
{
Annotation annotation = _processor.Annotate(article);
HashSet <string> persons = NLPCoreHelper.GetPersons(annotation);
HashSet <string> organizations = NLPCoreHelper.GetOrganizations(annotation);
HashSet <string> locations = NLPCoreHelper.GetLocation(annotation);
HashSet <string> sportSpecificWords = NLPCoreHelper.GetOccurence(annotation, _sportSpecificWords);
double[] personFeatures = PersonOccurenceDatabase.GetFeatures(persons);
double[] organizationFeatures = OrganizationOccurenceDatabase.GetFeatures(organizations);
double[] locationFeatures = LocationsOccurenceDatabase.GetFeatures(locations);
double[] specificFeatures = SportSpecificWordsOccurenceDatabase.GetFeatures(sportSpecificWords);
double[] features = new double[personFeatures.Length + organizationFeatures.Length + locationFeatures.Length + specificFeatures.Length];
Array.Copy(personFeatures, 0, features, 0, personFeatures.Length);
Array.Copy(organizationFeatures, 0, features, personFeatures.Length, organizationFeatures.Length);
Array.Copy(locationFeatures, 0, features, personFeatures.Length + organizationFeatures.Length, locationFeatures.Length);
Array.Copy(specificFeatures, 0, features, personFeatures.Length + organizationFeatures.Length + locationFeatures.Length, specificFeatures.Length);
for (int i = 0; i < features.Length; i++)
{
features[i] -= 0.5;
}
return(features);
}
public Marker(NLPProcessor processor, string modelPath)
{
_processor = processor;
_modelPath = modelPath;
Annotation sportSpecificWords = _processor.Annotate(File.ReadAllText(modelPath + "specificwords.txt"));
_sportSpecificWords = NLPCoreHelper.GetLemmas(sportSpecificWords);
File.Delete(modelPath + "specificwords.txt");
StringBuilder builder = new StringBuilder();
foreach (string word in _sportSpecificWords)
{
builder.AppendLine(word);
}
File.WriteAllText(modelPath + "specificwords.txt", builder.ToString());
PersonOccurenceDatabase = new WordOccurenceDatabase();
OrganizationOccurenceDatabase = new WordOccurenceDatabase();
LocationsOccurenceDatabase = new WordOccurenceDatabase();
SportSpecificWordsOccurenceDatabase = new WordOccurenceDatabase();
_categoriesCount = Enum.GetValues(typeof(SportCategory)).Length;
int hiddenLayerCount = (10 * _categoriesCount) / 3;
_classifier = new ActivationNetwork(new SigmoidFunction(), 4 * _categoriesCount, hiddenLayerCount, _categoriesCount);
_teacher = new BackPropagationLearning(_classifier);
Array values = Enum.GetValues(typeof(SportCategory));
_categoryIndex = new Dictionary <SportCategory, int>();
_indexCategory = new Dictionary <int, SportCategory>();
for (int index = 0; index < values.Length; index++)
{
SportCategory category = (SportCategory)values.GetValue(index);
_categoryIndex.Add(category, index);
_indexCategory.Add(index, category);
}
}
