@SuppressWarnings("unchecked")
public DictionaryCatalog(
Dictionary dictionary,
DictionaryFileType fileType,
Class desiredDictionaryFileType,
Map<String, Param> params)
throws JWNLException {
this.files = new EnumMap<POS, E>(POS.class);
this.dictionary = dictionary;
this.fileType = fileType;
if (!params.containsKey(DICTIONARY_PATH_KEY)) {
throw new JWNLException(
dictionary.getMessages().resolveMessage("DICTIONARY_EXCEPTION_052", DICTIONARY_PATH_KEY));
}
String path = params.get(DICTIONARY_PATH_KEY).getValue();
if (!params.containsKey(DICTIONARY_FILE_TYPE_KEY)) {
throw new JWNLException(
dictionary
.getMessages()
.resolveMessage("DICTIONARY_EXCEPTION_052", DICTIONARY_FILE_TYPE_KEY));
}
try {
Class fileClass;
try {
fileClass = Class.forName(params.get(DICTIONARY_FILE_TYPE_KEY).getValue());
if (!desiredDictionaryFileType.isAssignableFrom(fileClass)) {
throw new JWNLException(
dictionary
.getMessages()
.resolveMessage(
"DICTIONARY_EXCEPTION_003",
new Object[] {fileClass, desiredDictionaryFileType.getCanonicalName()}));
}
} catch (ClassNotFoundException e) {
throw new JWNLException(
dictionary.getMessages().resolveMessage("DICTIONARY_EXCEPTION_002"), e);
}
DictionaryFileFactory<E> factory =
(DictionaryFileFactory<E>) params.get(DICTIONARY_FILE_TYPE_KEY).create();
for (POS pos : POS.getAllPOS()) {
E file = factory.newInstance(dictionary, path, pos, fileType);
files.put(file.getPOS(), file);
}
} catch (JWNLException e) {
throw new JWNLException(
dictionary.getMessages().resolveMessage("DICTIONARY_EXCEPTION_018", fileType), e);
}
}
public Synset(Dictionary dictionary, POS pos) throws JWNLException {
super(dictionary);
if (null == pos) {
throw new IllegalArgumentException(JWNL.resolveMessage("DICTIONARY_EXCEPTION_041"));
}
this.pos = pos;
pointers = new PointerList();
words = new WordList();
if (null != dictionary && dictionary.isEditable()) {
dictionary.addSynset(this);
}
}
@Override
public void setDictionary(Dictionary dictionary) throws JWNLException {
if (dictionary != this.dictionary) {
if (null != this.dictionary) {
Dictionary old = this.dictionary;
this.dictionary = dictionary;
old.removeElement(this);
}
super.setDictionary(dictionary);
if (null != dictionary) {
dictionary.addElement(this);
}
}
}
private static Set<Long> findNounHypernyms(Dictionary dic) throws SMatchException {
log.info("Creating noun hypernyms array...");
try {
Set<Long> keys = new HashSet<>();
int count = 0;
Iterator<Synset> it = dic.getSynsetIterator(POS.NOUN);
while (it.hasNext()) {
count++;
if (0 == count % 10000) {
log.debug("noun hypernyms: " + count);
}
Synset source = it.next();
long sourceOffset = source.getOffset();
traverseTreeMG(keys, PointerUtils.getHypernymTree(source), sourceOffset);
traverseTreeMG(keys, PointerUtils.getInheritedHolonyms(source), sourceOffset);
traverseTreeMG(keys, PointerUtils.getInheritedMemberHolonyms(source), sourceOffset);
traverseTreeMG(keys, PointerUtils.getInheritedPartHolonyms(source), sourceOffset);
traverseTreeMG(keys, PointerUtils.getInheritedSubstanceHolonyms(source), sourceOffset);
traverseListMG(keys, PointerUtils.getHolonyms(source), sourceOffset);
traverseListMG(keys, PointerUtils.getMemberHolonyms(source), sourceOffset);
traverseListMG(keys, PointerUtils.getPartHolonyms(source), sourceOffset);
traverseListMG(keys, PointerUtils.getSubstanceHolonyms(source), sourceOffset);
}
log.info("Noun hypernyms: " + keys.size());
return keys;
} catch (JWNLException e) {
throw new SMatchException(e.getClass().getSimpleName() + ": " + e.getMessage(), e);
}
}
/* (non-Javadoc)
* @see ac.biu.nlp.nlp.instruments.dictionary.wordnet.jwnl.PartialJwnlDictionary#getIndexWord(java.lang.String, ac.biu.nlp.nlp.instruments.dictionary.wordnet.WordNetPartOfSpeech)
*/
protected IndexWord getIndexWord(String lemma, WordNetPartOfSpeech partOfSpeech)
throws JWNLException {
return doNotProcessThisWord(lemma)
? null
: extJwnlRealDictionary.lookupIndexWord(
ExtJwnlUtils.getJwnlPartOfSpeec(partOfSpeech), lemma);
}
private static Set<Long> findAdverbAntonyms(Dictionary dic) throws SMatchException {
log.info("Creating adverb antonyms array...");
try {
Set<Long> keys = new HashSet<>();
int count = 0;
Iterator<Synset> it = dic.getSynsetIterator(POS.ADVERB);
while (it.hasNext()) {
count++;
if (0 == count % 1000) {
log.debug("adverb antonyms: " + count);
}
Synset source = it.next();
long sourceOffset = source.getOffset();
List<Pointer> pointers = source.getPointers(PointerType.ANTONYM);
for (Pointer ptr : pointers) {
long targetOffset = ptr.getTargetOffset();
long key;
if (targetOffset > sourceOffset) {
key = (targetOffset << 32) + sourceOffset;
} else {
key = (sourceOffset << 32) + targetOffset;
}
keys.add(key);
}
}
log.info("Adverbs antonyms: " + keys.size());
return keys;
} catch (JWNLException e) {
throw new SMatchException(e.getClass().getSimpleName() + ": " + e.getMessage(), e);
}
}
private static Set<Long> findNominalizations(Dictionary dic) throws SMatchException {
log.info("Creating nominalizations array...");
try {
Set<Long> keys = new HashSet<>();
int count = 0;
Iterator<Synset> it = dic.getSynsetIterator(POS.VERB);
while (it.hasNext()) {
count++;
if (0 == count % 1000) {
log.debug("nominalizations: " + count);
}
Synset source = it.next();
List<Pointer> pointers = source.getPointers(PointerType.DERIVATION);
for (Pointer pointer : pointers) {
if (POS.NOUN.equals(pointer.getTargetPOS())) {
long targetOffset = pointer.getTargetOffset();
long key = (source.getOffset() << 32) + targetOffset;
keys.add(key);
}
}
}
log.info("Nominalizations: " + keys.size());
return keys;
} catch (JWNLException e) {
throw new SMatchException(e.getClass().getSimpleName() + ": " + e.getMessage(), e);
}
}
/**
* Ctor
*
* @param wnDictionaryDir
* @throws WordNetInitializationException
*/
public ExtJwnlDictionary(File wnDictionaryDir) throws WordNetInitializationException {
if (wnDictionaryDir == null) throw new WordNetInitializationException("null WN directory");
if (!wnDictionaryDir.exists())
throw new WordNetInitializationException(
"given WN directory " + wnDictionaryDir + " doesn't exist");
File propsFile = null;
try {
propsFile = ExtJwnlDictionaryInitializer.makePropsFile(wnDictionaryDir);
this.extJwnlRealDictionary = Dictionary.getInstance(new FileInputStream(propsFile));
} catch (FileNotFoundException e) {
throw new WordNetInitializationException("props file " + propsFile + " not found", e);
} catch (net.sf.extjwnl.JWNLException e) {
throw new WordNetInitializationException(
"error instantiating a Dictionary with " + propsFile, e);
}
}
/**
* Return all {@link Synset}s of a particular POS
*
* @param pos
* @return
* @throws WordNetException
*/
public List<Synset> getAllWords(WordNetPartOfSpeech pos) throws WordNetException {
List<Synset> synsets = new ArrayList<Synset>();
try {
@SuppressWarnings("rawtypes")
Iterator iter =
extJwnlRealDictionary.getIndexWordIterator(ExtJwnlUtils.getJwnlPartOfSpeec(pos));
while (iter.hasNext()) {
IndexWord indexWord = (IndexWord) iter.next();
synsets.addAll(indexWordToList(indexWord));
}
} catch (JWNLException e) {
throw new WordNetException("looking for all <" + pos + ">s failed. See nested exception", e);
}
return synsets;
}
/*
* (non-Javadoc)
* @see ac.biu.nlp.nlp.instruments.dictionary.wordnet.Dictionary#getSynsetsOf(java.lang.String, ac.biu.nlp.nlp.instruments.dictionary.wordnet.WordNetPartOfSpeech)
*/
public Set<Synset> getSynsetsOf(String lemma, WordNetPartOfSpeech partOfSpeech)
throws WordNetException {
if (doNotProcessThisWord(lemma)) return new LinkedHashSet<Synset>();
else {
try {
IndexWord indexWord =
extJwnlRealDictionary.lookupIndexWord(
ExtJwnlUtils.getJwnlPartOfSpeec(partOfSpeech), lemma);
return indexWordToSet(indexWord);
} catch (JWNLException e) {
throw new WordNetException(
"looking for word <"
+ lemma
+ "> with part of speech: "
+ partOfSpeech.toString()
+ " failed. See nested exception",
e);
}
}
}
private static Set<Long> findAdjectiveAntonyms(Dictionary dic) throws SMatchException {
log.info("Creating adjective antonyms array...");
try {
Set<Long> keys = new HashSet<>();
int count = 0;
Iterator<Synset> it = dic.getSynsetIterator(POS.ADJECTIVE);
while (it.hasNext()) {
count++;
if (0 == count % 1000) {
log.debug("adjective antonyms: " + count);
}
Synset current = it.next();
traverseTree(keys, PointerUtils.getExtendedAntonyms(current), current.getOffset());
traverseListSym(keys, PointerUtils.getAntonyms(current), current.getOffset());
}
log.info("Adjective antonyms: " + keys.size());
return keys;
} catch (JWNLException e) {
throw new SMatchException(e.getClass().getSimpleName() + ": " + e.getMessage(), e);
}
}
/*
* (non-Javadoc)
* @see ac.biu.nlp.nlp.instruments.dictionary.wordnet.Dictionary#getSortedSynsetOf(java.lang.String)
*/
public Map<WordNetPartOfSpeech, List<Synset>> getSortedSynsetOf(String lemma)
throws WordNetException {
Map<WordNetPartOfSpeech, List<Synset>> ret =
new LinkedHashMap<WordNetPartOfSpeech, List<Synset>>();
if (doNotProcessThisWord(lemma)) ;
else {
try {
IndexWordSet indexWordSet = extJwnlRealDictionary.lookupAllIndexWords(lemma);
if (indexWordSet != null) {
for (WordNetPartOfSpeech partOfSpeech : WordNetPartOfSpeech.values()) {
POS pos = ExtJwnlUtils.getJwnlPartOfSpeec(partOfSpeech);
if (indexWordSet.getIndexWord(pos) != null)
ret.put(partOfSpeech, indexWordToList(indexWordSet.getIndexWord(pos)));
}
}
} catch (JWNLException e) {
throw new WordNetException(
"looking for lemma <" + lemma + "> failed. See nested exception", e);
}
}
return ret;
}
/**
* Provides a WordNET dictionary
*
* @return wordnet
*/
public static Dictionary setupWordNet() {
// set up properties file
String propsFile = "file_properties.xml";
FileInputStream properties = null;
try {
properties = new FileInputStream(propsFile);
} catch (FileNotFoundException e1) {
e1.printStackTrace();
}
// create a dictionary and run the analytics
try {
// run
if (dictionary == null) {
// new style, instance dictionary
dictionary = Dictionary.getInstance(properties);
}
} catch (Exception e) {
e.printStackTrace();
System.exit(-1);
}
return dictionary;
}
private static Set<Long> findNounAntonyms(Dictionary dic) throws SMatchException {
log.info("Creating noun antonyms array...");
try {
Set<Long> keys = new HashSet<>();
int count = 0;
Iterator<Synset> it = dic.getSynsetIterator(POS.NOUN);
while (it.hasNext()) {
count++;
if (0 == count % 10000) {
log.debug("noun antonyms: " + count);
}
Synset source = it.next();
cartPr(keys, source.getPointers(PointerType.PART_MERONYM));
cartPr(keys, source.getPointers(PointerType.SUBSTANCE_MERONYM));
cartPr(keys, source.getPointers(PointerType.MEMBER_MERONYM));
}
log.info("Noun antonyms: " + keys.size());
return keys;
} catch (JWNLException e) {
throw new SMatchException(e.getClass().getSimpleName() + ": " + e.getMessage(), e);
}
}
public void run(String sentense, String word) throws JWNLException {
WORD = dictionary.lookupIndexWord(POS.NOUN, word);
leskImplementor.disambiguate(sentense, WORD);
}
private void readObject(java.io.ObjectInputStream in) throws IOException, ClassNotFoundException {
in.defaultReadObject();
dictionary = Dictionary.getRestoreDictionary();
}