/// <summary>
/// Constructor
/// </summary>
/// <param name="nouns">Nouns to get predicates trees for (null for all)</param>
/// <param name="nomBankEngine">NomBank engine to draw training predicate trees from</param>
/// <param name="instanceFilter">Instance filter to apply</param>
/// <param name="sections">TreeBank sections to draw instances from</param>
public NomBankPredicateTreeInstanceProvider(Set <string> nouns,
NomBankEngine nomBankEngine,
InstanceFilterDelegate instanceFilter,
Set <int> sections)
: base(nomBankEngine, instanceFilter, sections)
{
Nouns = nouns;
_filteredNodes = new List <NomBankNode>();
}
/// <summary>
/// Constructor
/// </summary>
/// <param name="labelPredicates">Whether or not to label predicates</param>
/// <param name="labelSupportVerbs">Whether or not to label support verbs</param>
/// <param name="autoParsedNomBankEngine">Automatically parsed version of NomBank, from which to draw parse tree structure</param>
/// <param name="goldParsedNomBankEngine">Gold-parsed version of NomBank, from which to draw predicate and support verb labels. This is
/// necessary because auto-parsed versions of NomBank don't include annotations for instances whose arguments cannot all be minimally
/// subsumed.</param>
/// <param name="instanceFilter">Instance filter to apply</param>
/// <param name="sections">TreeBank sections to draw instances from (null for all sections)</param>
public NomBankParseTreeInstanceProvider(bool labelPredicates,
bool labelSupportVerbs,
NomBankEngine autoParsedNomBankEngine,
NomBankEngine goldParsedNomBankEngine,
InstanceFilterDelegate instanceFilter,
Set <int> sections)
: base(autoParsedNomBankEngine, instanceFilter, sections)
{
_labelPredicates = labelPredicates;
_labelSupportVerbs = labelSupportVerbs;
_goldParsedNomBankEngine = goldParsedNomBankEngine;
}
/// <summary>
/// Constructor
/// </summary>
public TestForm()
{
InitializeComponent();
// initialize engine
string root = Directory.GetDirectoryRoot(".");
_nomBankEngine = new NomBankEngine(root + @"NLP\Resources\PennTreeBank_3\parsed\mrg\wsj",
root + @"NLP\Resources\nombank.1.0\nombank.1.0",
root + @"NLP\Resources\nombank.1.0\frames",
root + @"NLP\Resources\nombank.1.0\nombank-morph.dict.1.0",
root + @"NLP\Resources\nombank.1.0\nomlex-plus.1.0",
root + @"NLP\Resources\Indexes\nombank_index");
// populate noun box
foreach (string noun in _nomBankEngine.AllNouns)
{
nounCombo.Items.Add(noun);
}
// populate class box
foreach (string nlClass in _nomBankEngine.NomLexEngine.Classes)
{
classCombo.Items.Add(nlClass);
}
// populate NomLex noun box
List <string> nomLexNouns = new List <string>(_nomBankEngine.NomLexEngine.Nouns);
nomLexNouns.Sort();
foreach (string nomLexNoun in nomLexNouns)
{
nomLexNounCombo.Items.Add(nomLexNoun);
}
foreach (NomBankNodeLabel.NodeType t in Enum.GetValues(typeof(NomBankNodeLabel.NodeType)))
{
nodeTypeCombo.Items.Add(t);
}
foreach (NomBankNodeLabel.NodeFeature f in Enum.GetValues(typeof(NomBankNodeLabel.NodeFeature)))
{
nodeFeatureCombo.Items.Add(f);
}
}
/// <summary>
/// Tests the current node
/// </summary>
/// <param name="nomBankEngine">NomBankEngine to use in testing</param>
public void Test(NomBankEngine nomBankEngine)
{
// check predicate node
if (IsPredicate && !nomBankEngine.TokenIsMarkable(MrgFile, SentenceNumber, TokenNumber))
{
throw new Exception("Invalid markable token entry");
}
// check support verbs
if (IsSupportVerb && !nomBankEngine.TokenIsSupportVerb(MrgFile, SentenceNumber, TokenNumber))
{
throw new Exception("Support verb mismatch");
}
// check morphological variant
if (IsRoot && !nomBankEngine.ContainsMorphologicalVariantOf(Information.Noun))
{
throw new Exception("Noun does not exist in section");
}
// test children nodes
IEnumerator <TreeBankNode> childEnum = Children;
while (childEnum.MoveNext())
{
(childEnum.Current as NomBankNode).Test(nomBankEngine);
}
RoleSet roleSet = (Root as NomBankNode).RoleSet;
foreach (Role role in roleSet)
{
if (roleSet.Get(role.Number).Description != role.Description)
{
throw new Exception();
}
}
}
/// <summary>
/// Gets next training instance for models build over NomBank
/// </summary>
/// <returns>Next training instance</returns>
public override ClassifiableEntity GetNextInstance()
{
// try to move to next node
while (!_nodes.MoveNext())
{
NomBankEngine nomBankEngine = TreeBankEngine as NomBankEngine;
// try to move to next NounInfo
while (!MoveToNextValidNounInfo(ref _nounInfo))
{
// try to move to next noun...if there are none we're done
if (!_nouns.MoveNext())
{
return(null);
}
// start before first NounInfo for current noun
_nounInfo = nomBankEngine.GetNounInfo(_nouns.Current).GetEnumerator();
}
// filter all nodes in the tree, keeping the ones that pass
NomBankNode root = nomBankEngine.GetNomBankTree(_nounInfo.Current);
_filteredNodes.Clear(); // reuse node collection for better memory usage
foreach (NomBankNode n in root.AllNodes)
{
if (Filter(n))
{
_filteredNodes.Add(n);
}
}
_nodes = _filteredNodes.GetEnumerator();
}
return(_nodes.Current);
}
