/**
* given an AgentID, find the text span associated with this Agent and highlight the appropriate
* areas in the raw text panel
*
* @param agentRef the AgentID
*/
public void highlightText(Annotation an, Color highlightTextColor) {
if (an != null) {
// highlight the annotation
// also define color to the highlight
tvPanel.highlightSpan(
an.getStartOffset(), an.getEndOffset() - an.getStartOffset(), highlightTextColor);
}
}
public static Annotation getHeadVerb(
Annotation verb, Map<String, AnnotationSet> annotations, String text) {
AnnotationSet verbs = annotations.get(Constants.POS).getContained(verb);
if (verbs == null || verbs.size() < 1) return verb;
Annotation result = verbs.getFirst();
for (Annotation v : verbs) {
if (!SyntaxUtils.isPreposition(v.getType())) {
result = v;
}
}
return result;
}
@Override
public String produceValue(
Annotation np1, Annotation np2, Document doc, Map<Feature, String> featVector) {
if (!ProperName.getValue(np1, doc) || !ProperName.getValue(np2, doc)) return NA;
Annotation ne1 = (Annotation) np1.getProperty(Property.LINKED_PROPER_NAME);
Annotation ne2 = (Annotation) np2.getProperty(Property.LINKED_PROPER_NAME);
String[] infW1 = InfWords.getValue(ne1, doc);
String[] infW2 = InfWords.getValue(ne2, doc);
if (infW1 == null || infW2 == null || infW1.length < 1 || infW2.length < 1) return INCOMPATIBLE;
if (Utils.isAnySubset(infW1, infW2)) return COMPATIBLE;
else return INCOMPATIBLE;
}
public Component getListCellRendererComponent(
JList list, Object value, int index, boolean isSelected, boolean cellHasFocus) {
if (!(value instanceof Annotation)) throw new RuntimeException("Invalid value in opinion list");
// System.err.println("Rendering..."+value);
Annotation a = (Annotation) value;
StringBuilder s = new StringBuilder(a.getType());
s.append(" - ");
Map<String, String> features = a.getFeatures();
for (String k : features.keySet()) {
s.append(k).append(": ").append(features.get(k)).append(" ");
}
setText(s.toString());
if (isSelected) {
setBackground(list.getSelectionBackground());
setForeground(list.getSelectionForeground());
} else {
setBackground(list.getBackground());
setForeground(list.getForeground());
}
setEnabled(list.isEnabled());
setFont(list.getFont());
setOpaque(true);
return this;
}
@Override
public Object produceValue(Annotation np, Document doc) {
// Get the sentence annotations
AnnotationSet par = doc.getAnnotationSet(Constants.PAR);
AnnotationSet nps = doc.getAnnotationSet(Constants.NP);
for (Annotation p : par) {
int num;
if (Constants.PAR_NUMS_UNAVAILABLE) {
num = 0;
} else {
num = Integer.parseInt(p.getAttribute("parNum"));
}
AnnotationSet enclosed = nps.getContained(p);
for (Annotation e : enclosed) {
e.setProperty(this, num);
}
}
// Make sure that all annotations have an associated PARNUM
for (Annotation n : nps) {
if (n.getProperty(this) == null) {
AnnotationSet o = par.getOverlapping(0, n.getEndOffset());
if (o == null || o.size() < 1) {
n.setProperty(this, 0);
} else {
Annotation p = o.getLast();
int num = Integer.parseInt(p.getAttribute("parNum")); // = 0;
n.setProperty(this, num);
}
}
}
if (np.getProperty(this) == null) {
AnnotationSet o = par.getOverlapping(0, np.getEndOffset());
if (o == null || o.size() < 1) {
np.setProperty(this, 0);
} else {
Annotation p = o.getLast();
int num = Integer.parseInt(p.getAttribute("parNum"));
np.setProperty(this, num);
}
}
return np.getProperty(this);
}
@Override
public void run(Document doc, String annSetNames[]) {
String tagChunk = currentConfig.getTagChunk();
String listDir = currentConfig.getTagChunkLists();
AnnotationSet namedEntities = new AnnotationSet(annSetNames[0]);
// get the sentences from the input
AnnotationSet sentSet = doc.getAnnotationSet(Constants.SENT);
// get the tokens from each sentence
AnnotationSet tokenSet = doc.getAnnotationSet(Constants.TOKEN);
// Read in the text from the raw file
String text = doc.getText();
Iterator<Annotation> sents = sentSet.iterator();
ArrayList<String> lines = new ArrayList<String>();
ArrayList<Vector<Annotation>> tokenList = new ArrayList<Vector<Annotation>>();
while (sents.hasNext()) {
Vector<Annotation> annVector = new Vector<Annotation>();
Annotation sent = sents.next();
int sentStart = sent.getStartOffset();
int sentEnd = sent.getEndOffset();
String sentText = Annotation.getAnnotText(sent, text);
AnnotationSet sentTokens = tokenSet.getContained(sentStart, sentEnd);
// gather all sentences to tag
if (!sentText.matches("\\W+")) {
StringBuilder tmp = new StringBuilder();
for (Annotation a : sentTokens) {
tmp.append(Annotation.getAnnotTextClean(a, text)).append(" ");
annVector.add(a);
}
lines.add(tmp.toString());
tokenList.add(annVector);
}
}
// write out a tmp file that contains the words to be tagged
File tmpFile = new File(doc.getRootDir(), "tmp.ner");
try {
tmpFile.deleteOnExit();
FileWriter fw = new FileWriter(tmpFile);
BufferedWriter bw = new BufferedWriter(fw);
for (String l : lines) {
// System.out.println(l);
bw.write(l + "\n");
}
bw.close();
fw.close();
} catch (IOException ioe) {
ioe.printStackTrace();
}
// run the tagger
String command =
tagChunk
+ " -predict . "
+ modelDir
+ Utils.SEPARATOR
+ models[0]
+ " "
+ tmpFile.getAbsolutePath()
+ " "
+ listDir;
// collect the results
ArrayList<String> results;
int i = 0;
try {
results = Utils.runExternalCaptureOutput(command);
Annotation current = null;
for (String l : results) {
Vector<Annotation> annVector = tokenList.get(i);
// get rid of these extraneous tags
l = l.replace("_O-O", "");
String[] tokens = l.split(" ");
// System.out.println(l);
int j = 0;
int underscore;
int nes = 1;
String tag;
for (String t : tokens) {
underscore = t.lastIndexOf('_');
tag = t.substring(underscore + 1, t.length());
Annotation a = annVector.get(j);
// System.out.print(Utils.getAnnotTextClean(a, text) + "_" + tag + " ");
if (tag.equals("B-O")) {
j++;
if (current != null) {
namedEntities.add(current);
// System.out.println("NE Found: " + Utils.getAnnotTextClean(current, text) + ":" +
// current.getType());
nes++;
current = null;
}
continue;
}
String entityType = tag.substring(tag.indexOf("-") + 1, tag.length());
if (entityType.equals("ORG")) {
entityType = "ORGANIZATION";
} else if (entityType.equals("LOC")) {
entityType = "LOCATION";
} else if (entityType.equals("PER")) {
entityType = "PERSON";
} else if (entityType.equals("VEH")) {
entityType = "VEHICLE";
}
if (tag.startsWith("B-")) {
if (current != null) {
namedEntities.add(current);
nes++;
current = null;
// System.out.println("NE Found: " + Utils.getAnnotTextClean(current, text));
}
current = new Annotation(nes, a.getStartOffset(), a.getEndOffset(), entityType);
} else if (tag.startsWith("I-")) {
if (current != null) {
current.setEndOffset(a.getEndOffset());
} else {
current = new Annotation(nes, a.getStartOffset(), a.getEndOffset(), entityType);
}
}
j++;
}
// System.out.println();
i++;
}
FileUtils.delete(tmpFile);
} catch (IOException e) {
throw new RuntimeException(e);
} catch (InterruptedException e) {
throw new RuntimeException(e);
}
addResultSet(doc, namedEntities);
}
public static Annotation getContainedProperName(Annotation np, Document doc) {
Annotation cont = ContainsProperName.getValue(np, doc);
if (cont.getId() < 0) return null;
else return cont;
}
public static boolean containsProperName(Annotation np, Document doc) {
Annotation cont = ContainsProperName.getValue(np, doc);
if (cont.getId() < 0) return false;
else return true;
}
public static boolean isAdverb(Annotation pos) {
final String[] ADV_POS = {"RB", "RBR", "RBS"};
return memberArray(pos.getType(), ADV_POS);
}
public static boolean isAdjective(Annotation pos) {
final String[] ADJ_POS = {"JJ", "JJR", "JJS"};
return memberArray(pos.getType(), ADJ_POS);
}
public static boolean isPossesivePronoun(Annotation pos) {
return pos.getType().equals("PRP$");
}
public static boolean isPredeterminer(Annotation pos) {
return pos.getType().equals("PDT");
}
public static boolean isProperNoun(Annotation pos) {
return pos.getType().startsWith("NNP");
}
public static boolean isCardinalNumber(Annotation pos) {
final String[] CN_POS = {"CC"};
return memberArray(pos.getType(), CN_POS);
}