/// <summary>
/// Extracts and returns all terms with the specified POS from a string.
/// </summary>
/// <param name="text">Source string</param>
/// <param name="pos">Part of speech to extract</param>
/// <returns>String consisting of only the terms with the specified POS</returns>
public static string ExtractPOS(string text, POSTaggerSpeechType pos)
{
Dictionary<POSTaggerSpeechType, List<string>> dict = new Dictionary<POSTaggerSpeechType, List<string>>();
foreach (POSTaggerSpeechType type in Enum.GetValues(typeof(POSTaggerSpeechType)))
{
dict.Add(type, new List<string>());
}
foreach (string term in text.Split())
{
string[] kvp = term.Split('_');
if (kvp.Length < 2)
continue;
if (kvp[1].Length == 0)
continue;
if (kvp[1][0] == 'N')
{
dict[POSTaggerSpeechType.Noun].Add(kvp[0]);
continue;
}
if (kvp[1][0] == 'V')
{
dict[POSTaggerSpeechType.Verb].Add(kvp[0]);
continue;
}
dict[POSTaggerSpeechType.Other].Add(kvp[0]);
}
return String.Join(" ", dict[pos]);
}
/// <summary>
/// Extracts and returns all terms with the specified POS from a string.
/// </summary>
/// <param name="text">Source string</param>
/// <param name="pos">Part of speech to extract</param>
/// <returns>String consisting of only the terms with the specified POS</returns>
public static string ExtractPOS(string text, POSTaggerSpeechType pos)
{
Dictionary <POSTaggerSpeechType, List <string> > dict = new Dictionary <POSTaggerSpeechType, List <string> >();
foreach (POSTaggerSpeechType type in Enum.GetValues(typeof(POSTaggerSpeechType)))
{
dict.Add(type, new List <string>());
}
foreach (string term in text.Split())
{
string[] kvp = term.Split('_');
if (kvp.Length < 2)
{
continue;
}
if (kvp[1].Length == 0)
{
continue;
}
if (kvp[1][0] == 'N')
{
dict[POSTaggerSpeechType.Noun].Add(kvp[0]);
continue;
}
if (kvp[1][0] == 'V')
{
dict[POSTaggerSpeechType.Verb].Add(kvp[0]);
continue;
}
dict[POSTaggerSpeechType.Other].Add(kvp[0]);
}
return(String.Join(" ", dict[pos]));
}
/// <summary>
/// Extracts and returns all terms with the specified POS from a TLArtifactsCollection.
/// </summary>
/// <param name="artifacts">List of artifacts</param>
/// <param name="pos">Part of speech to extract</param>
/// <param name="modelFile">Training model file location</param>
/// <returns>TLArtifactsCollection consisting of only the terms with the specified POS</returns>
public static TLArtifactsCollection Extract(TLArtifactsCollection artifacts, POSTaggerSpeechType pos, string modelFile)
{
TLArtifactsCollection extracted = new TLArtifactsCollection();
foreach (KeyValuePair<string, TLArtifact> artifactKVP in artifacts)
{
extracted.Add(artifactKVP.Key, ExtractArtifact(artifactKVP.Value, pos, modelFile));
}
return extracted;
}
/// <summary>
/// Extracts and returns all terms with the specified POS from a TLArtifactsCollection.
/// </summary>
/// <param name="artifacts">List of artifacts</param>
/// <param name="pos">Part of speech to extract</param>
/// <param name="modelFile">Training model file location</param>
/// <returns>TLArtifactsCollection consisting of only the terms with the specified POS</returns>
public static TLArtifactsCollection Extract(TLArtifactsCollection artifacts, POSTaggerSpeechType pos, string modelFile)
{
TLArtifactsCollection extracted = new TLArtifactsCollection();
foreach (KeyValuePair <string, TLArtifact> artifactKVP in artifacts)
{
extracted.Add(artifactKVP.Key, ExtractArtifact(artifactKVP.Value, pos, modelFile));
}
return(extracted);
}
/// <summary>
/// Extracts and returns all terms with the specified POS from a single TLArtifact.
/// </summary>
/// <param name="artifact">Single artifact</param>
/// <param name="pos">Part of speech to extract</param>
/// <param name="modelFile">Training model file location</param>
/// <returns>Single artifact consisting of only the terms with the specified POS</returns>
public static TLArtifact ExtractArtifact(TLArtifact artifact, POSTaggerSpeechType pos, string modelFile)
{
return new TLArtifact(artifact.Id, ExtractPOS(Tag(artifact.Text, modelFile), pos));
}
/// <summary>
/// Extracts and returns all terms with the specified POS from a single TLArtifact.
/// </summary>
/// <param name="artifact">Single artifact</param>
/// <param name="pos">Part of speech to extract</param>
/// <param name="modelFile">Training model file location</param>
/// <returns>Single artifact consisting of only the terms with the specified POS</returns>
public static TLArtifact ExtractArtifact(TLArtifact artifact, POSTaggerSpeechType pos, string modelFile)
{
return(new TLArtifact(artifact.Id, ExtractPOS(Tag(artifact.Text, modelFile), pos)));
}
