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);
}