public ArrayList <Annotation> ExtractEntities(string text, int offset)
{
mLogger.Debug("ExtractEntities", "Extracting entities ...");
mTokenizer.Text = text;
ArrayList <Annotation> annotations = new ArrayList <Annotation>();
ArrayList <string> tmp = new ArrayList <string>();
ArrayList <int> pos = new ArrayList <int>();
for (SimpleTokenizer.Enumerator e = mTokenizer.GetEnumerator(); e.MoveNext();)
{
tmp.Add(Normalize(e.Current));
pos.Add(e.CurrentTokenIdx);
}
string[] tokens = tmp.ToArray();
foreach (Gazetteer gazetteer in mGazetteers.Values)
{
if (gazetteer.mTerms.Count > 0)
{
int len;
for (int startIdx = 0; startIdx < tokens.Length; startIdx++)
{
if (gazetteer.Match(tokens, startIdx, out len))
{
ArrayList <ArrayList <Entity> > allPaths = new ArrayList <ArrayList <Entity> >();
GetPaths(0, new ArrayList <Entity>(new Entity[] { gazetteer.mUri }), allPaths);
foreach (ArrayList <Entity> path in allPaths)
{
string pathStr = "";
for (int i = path.Count - 1; i > 1; i--)
{
pathStr += GetLabel(path[i], mRdfStore) + "/";
}
pathStr = pathStr.TrimEnd('/');
annotations.Add(new Annotation(pos[startIdx] + offset, pos[startIdx + len - 1] + tokens[startIdx + len - 1].Length - 1 + offset, pathStr));
annotations.Last.Features.SetFeatureValue("gazUri", gazetteer.mUri);
if (path.Count >= 2)
{
annotations.Last.Features.SetFeatureValue("objUri", path[1].Uri);
annotations.Last.Features.SetFeatureValue("objLabel", GetLabel(path[1], mRdfStore));
}
}
}
}
}
}
return(annotations);
}
