Skip to content

Commit

Permalink
Release: 3.1.1.
Browse files Browse the repository at this point in the history
  • Loading branch information
@jdchoi77
jdchoi77 committed May 7, 2015
1 parent 10b2851 commit 5893df8
Show file tree
Hide file tree
Showing 28 changed files with 330 additions and 678 deletions.
10 changes: 10 additions & 0 deletions src/main/java/edu/emory/clir/clearnlp/component/mode/ner/NERFeatureExtractor.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -44,4 +44,14 @@ protected String getFeature(CommonFeatureToken token, NERState state, DEPNode no
default: return super.getFeature(token, state, node);
}
}

// @Override
// protected String[] getFeatures(CommonFeatureToken token, NERState state, DEPNode node)
// {
// switch (token.getField())
// {
// case as: return state.getAmbiguityClasses(node);
// default: return super.getFeatures(token, state, node);
// }
// }
}
49 changes: 48 additions & 1 deletion src/main/java/edu/emory/clir/clearnlp/component/mode/ner/NERState.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -55,10 +55,52 @@ public NERState(DEPTree tree, CFlag flag, PrefixTree<String,NERInfoSet> namedEnt
public void init(PrefixTree<String,NERInfoSet> namedEntityDictionary)
{
ne_dictionary = namedEntityDictionary;
info_list = ne_dictionary.getAll(d_tree.toNodeArray(), 1, DEPNode::getWordForm, true, false);
// info_list = ne_dictionary.getAll(d_tree.toNodeArray(), 1, DEPNode::getWordForm, true, false);
info_list = ne_dictionary.getAll(d_tree.toNodeArray(), 1, DEPNode::getLowerSimplifiedWordForm, true, false);
ambiguity_classes = getAmbiguityClasses();
}

// private void initAmbiguityClasses()
// {
// List<Set<String>> sets = IntStream.range(0, t_size).mapToObj(k -> new HashSet<String>()).collect(Collectors.toList());
// StringJoiner[] joiners = new StringJoiner[t_size];
// int i, j, size = info_list.size();
// ObjectIntIntTriple<NERInfoSet> t;
// String tag;
//
// for (i=1; i<t_size; i++)
// joiners[i] = new StringJoiner("-");
//
// for (i=0; i<size; i++)
// {
// t = info_list.get(i);
// tag = t.o.joinTags(StringConst.COLON);
//
// if (t.i1 == t.i2)
// joiners[t.i1].add(NERLib.toBILOUTag(BILOU.U, tag));
// else
// {
// joiners[t.i1].add(NERLib.toBILOUTag(BILOU.B, tag));
// joiners[t.i2].add(NERLib.toBILOUTag(BILOU.L, tag));
//
// for (j=t.i1+1; j<t.i2; j++)
// joiners[j].add(NERLib.toBILOUTag(BILOU.I, tag));
// }
//
// for (j=t.i1; j<=t.i2; j++)
// sets.get(j).addAll(t.o.getCategorySet());
// }
//
// ambiguity_class_set = new String[t_size][];
// ambiguity_classes = new String[t_size];
//
// for (i=1; i<t_size; i++)
// {
// ambiguity_classes [i] = joiners.length == 0 ? null : joiners[i].toString();
// ambiguity_class_set[i] = DSUtils.toArray(sets.get(i));
// }
// }

private String[] getAmbiguityClasses()
{
StringJoiner[] joiners = new StringJoiner[t_size];
Expand Down Expand Up @@ -116,6 +158,11 @@ public String getAmbiguityClass(DEPNode node)
return ambiguity_classes[node.getID()];
}

// public String[] getAmbiguityClasses(DEPNode node)
// {
// return ambiguity_class_set[node.getID()];
// }
//
// public String[] getCooccuranceFeatures(DEPNode node)
// {
// String[] categories = {"PER", "LOC", "ORG", "MISC"};
Expand Down
2 changes: 1 addition & 1 deletion src/main/java/edu/emory/clir/clearnlp/component/utils/NLPUtils.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -129,7 +129,7 @@ static public AbstractNERecognizer getNERecognizer(TLanguage language, String mo
@SuppressWarnings("unchecked")
static public PrefixTree<String,NERInfoSet> getNERDictionary(ObjectInputStream in)
{
BinUtils.LOG.info("Loading named entity dictionary.\n");
BinUtils.LOG.info("Loading named entity gazetteers.\n");
PrefixTree<String,NERInfoSet> tree = null;

try
Expand Down
68 changes: 68 additions & 0 deletions src/main/java/edu/emory/clir/clearnlp/constituent/CTLib.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,9 @@
package edu.emory.clir.clearnlp.constituent;

import java.util.List;
import java.util.Set;
import java.util.function.Predicate;
import java.util.regex.Pattern;

/**
* @author Jinho D. Choi ({@code jinho.choi@emory.edu})
Expand All @@ -35,4 +38,69 @@ static public String toForms(List<CTNode> tokens, int beginIndex, int endIndex,

return build.substring(delim.length());
}

static public Predicate<CTNode> matchC(String constituentTag)
{
return node -> node.isConstituentTag(constituentTag);
}

static public Predicate<CTNode> matchCo(String... constituentTags)
{
return node -> node.isConstituentTagAny(constituentTags);
}

static public Predicate<CTNode> matchCo(Set<String> constituentTags)
{
return node -> constituentTags.contains(node.getConstituentTag());
}

static public Predicate<CTNode> matchCp(String constituentPrefix)
{
return node -> node.getConstituentTag().startsWith(constituentPrefix);
}

static public Predicate<CTNode> matchCF(String constituentTag, String functionTag)
{
return node -> node.isConstituentTag(constituentTag) && node.hasFunctionTag(functionTag);
}

static public Predicate<CTNode> matchCFa(String constituentTag, String... functionTags)
{
return node -> node.isConstituentTag(constituentTag) && node.hasFunctionTagAll(functionTags);
}

static public Predicate<CTNode> matchCFo(String constituentTag, String... functionTags)
{
return node -> node.isConstituentTag(constituentTag) && node.hasFunctionTagAny(functionTags);
}

static public Predicate<CTNode> matchF(String functionTag)
{
return node -> node.hasFunctionTag(functionTag);
}

static public Predicate<CTNode> matchFa(String... functionTags)
{
return node -> node.hasFunctionTagAll(functionTags);
}

static public Predicate<CTNode> matchFo(String... functionTags)
{
return node -> node.hasFunctionTagAny(functionTags);
}

static public Predicate<CTNode> matchP(Pattern constituentPattern)
{
return node -> node.matchesConstituentTag(constituentPattern);
}

static public Predicate<CTNode> matchPFa(Pattern constituentPattern, String... functionTags)
{
return node -> node.matchesConstituentTag(constituentPattern) && node.hasFunctionTagAll(functionTags);
}

static public Predicate<CTNode> matchPFo(Pattern constituentPattern, String... functionTags)
{
return node -> node.matchesConstituentTag(constituentPattern) && node.hasFunctionTagAny(functionTags);
}
}
45 changes: 20 additions & 25 deletions src/main/java/edu/emory/clir/clearnlp/constituent/CTLibEn.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -17,14 +17,9 @@

import java.util.List;
import java.util.Set;
import java.util.function.Predicate;
import java.util.regex.Pattern;

import edu.emory.clir.clearnlp.constituent.matcher.CTNodeMatcher;
import edu.emory.clir.clearnlp.constituent.matcher.CTNodeMatcherC;
import edu.emory.clir.clearnlp.constituent.matcher.CTNodeMatcherCF;
import edu.emory.clir.clearnlp.constituent.matcher.CTNodeMatcherF;
import edu.emory.clir.clearnlp.constituent.matcher.CTNodeMatcherPrefix;
import edu.emory.clir.clearnlp.constituent.matcher.CTNodeMatcherSet;
import edu.emory.clir.clearnlp.pos.POSLibEn;
import edu.emory.clir.clearnlp.pos.POSTagEn;
import edu.emory.clir.clearnlp.util.DSUtils;
Expand All @@ -41,26 +36,26 @@ public class CTLibEn extends CTLib implements CTTagEn, POSTagEn
{
static final public Pattern P_PASSIVE_NULL = PatternUtils.createClosedORPattern("\\*","\\*-\\d+");

static final public CTNodeMatcher M_NP = new CTNodeMatcherC(C_NP);
static final public CTNodeMatcher M_VP = new CTNodeMatcherC(C_VP);
static final public CTNodeMatcher M_QP = new CTNodeMatcherC(C_QP);
static final public CTNodeMatcher M_ADVP = new CTNodeMatcherC(C_ADVP);
static final public CTNodeMatcher M_SBAR = new CTNodeMatcherC(C_SBAR);
static final public CTNodeMatcher M_EDITED = new CTNodeMatcherC(C_EDITED);
static final public Predicate<CTNode> M_NP = CTLib.matchC(C_NP);
static final public Predicate<CTNode> M_VP = CTLib.matchC(C_VP);
static final public Predicate<CTNode> M_QP = CTLib.matchC(C_QP);
static final public Predicate<CTNode> M_ADVP = CTLib.matchC(C_ADVP);
static final public Predicate<CTNode> M_SBAR = CTLib.matchC(C_SBAR);
static final public Predicate<CTNode> M_EDITED = CTLib.matchC(C_EDITED);

static final public CTNodeMatcher M_NOM = new CTNodeMatcherF(F_NOM);
static final public CTNodeMatcher M_PRD = new CTNodeMatcherF(F_PRD);
static final public Predicate<CTNode> M_NOM = CTLib.matchF(F_NOM);
static final public Predicate<CTNode> M_PRD = CTLib.matchF(F_PRD);

static final public CTNodeMatcher M_NP_SBJ = new CTNodeMatcherCF(C_NP, F_SBJ);
static final public Predicate<CTNode> M_NP_SBJ = CTLib.matchCF(C_NP, F_SBJ);

static final public CTNodeMatcher M_NNx = new CTNodeMatcherPrefix(POS_NN);
static final public CTNodeMatcher M_VBx = new CTNodeMatcherPrefix(POS_VB);
static final public CTNodeMatcher M_WHx = new CTNodeMatcherPrefix("WH");
static final public CTNodeMatcher M_SBARx = new CTNodeMatcherPrefix(C_SBAR);
static final public Predicate<CTNode> M_NNx = CTLib.matchCp(POS_NN);
static final public Predicate<CTNode> M_VBx = CTLib.matchCp(POS_VB);
static final public Predicate<CTNode> M_WHx = CTLib.matchCp("WH");
static final public Predicate<CTNode> M_SBARx = CTLib.matchCp(C_SBAR);

static final public CTNodeMatcher M_NP_NML = new CTNodeMatcherSet(DSUtils.toHashSet(C_NP, C_NML));
static final public CTNodeMatcher M_VBD_VBN = new CTNodeMatcherSet(DSUtils.toHashSet(POS_VBD, POS_VBN));
static final public CTNodeMatcher M_VP_RRC_UCP = new CTNodeMatcherSet(DSUtils.toHashSet(C_VP, C_RRC, C_UCP));
static final public Predicate<CTNode> M_NP_NML = CTLib.matchCo(DSUtils.toHashSet(C_NP, C_NML));
static final public Predicate<CTNode> M_VBD_VBN = CTLib.matchCo(DSUtils.toHashSet(POS_VBD, POS_VBN));
static final public Predicate<CTNode> M_VP_RRC_UCP = CTLib.matchCo(DSUtils.toHashSet(C_VP, C_RRC, C_UCP));

static final private Set<String> S_LGS_PHRASE = DSUtils.toHashSet(C_PP, C_SBAR);
static final private Set<String> S_MAIN_CLAUSE = DSUtils.toHashSet(C_S, C_SQ, C_SINV);
Expand Down Expand Up @@ -466,7 +461,7 @@ static public boolean isWhPhraseLink(CTNode node)

static public boolean isWhPhrase(CTNode node)
{
return M_WHx.matches(node);
return M_WHx.test(node);
}

static public boolean isEditedPhrase(CTNode node)
Expand All @@ -485,9 +480,9 @@ static public boolean isRNR(CTNode node)
return node.getWordForm().startsWith(E_RNR);
}

static public CTNode getNode(CTNode node, CTNodeMatcher matcher, boolean recursive)
static public CTNode getNode(CTNode node, Predicate<CTNode> matcher, boolean recursive)
{
if (matcher.matches(node))
if (matcher.test(node))
return node;

if (recursive && node.getChildrenSize() == 1)
Expand Down
Loading

0 comments on commit 5893df8

Please sign in to comment.