public XBarGrammarProjection(BinaryGrammar bg, UnaryGrammar ug) {
Map<BinaryRule, BinaryRule> binaryRules = new HashMap<BinaryRule, BinaryRule>();
Map<UnaryRule, UnaryRule> unaryRules = new HashMap<UnaryRule, UnaryRule>();
sourceUG = ug;
sourceBG = bg;
sourceNumberer = Numberer.getGlobalNumberer(bg.stateSpace());
targetNumberer = Numberer.getGlobalNumberer(bg.stateSpace() + "-xbar");
projection = new int[sourceNumberer.total()];
scanStates(sourceNumberer, targetNumberer);
targetBG = new BinaryGrammar(targetNumberer.total(), bg.stateSpace() + "-xbar");
targetUG = new UnaryGrammar(targetNumberer.total());
for (Iterator<BinaryRule> brI = bg.iterator(); brI.hasNext(); ) {
BinaryRule rule = projectBinaryRule(brI.next());
Rule old = binaryRules.get(rule);
if (old == null || rule.score > old.score) {
binaryRules.put(rule, rule);
}
}
for (BinaryRule br : binaryRules.keySet()) {
targetBG.addRule(br);
// System.out.println("BR: "+targetNumberer.object(br.parent)+" ->
// "+targetNumberer.object(br.leftChild)+" "+targetNumberer.object(br.rightChild)+" %%
// "+br.score);
}
targetBG.splitRules();
for (int parent = 0; parent < sourceNumberer.total(); parent++) {
for (Iterator<UnaryRule> urI = ug.ruleIteratorByParent(parent); urI.hasNext(); ) {
UnaryRule sourceRule = urI.next();
UnaryRule rule = projectUnaryRule(sourceRule);
Rule old = unaryRules.get(rule);
if (old == null || rule.score > old.score) {
unaryRules.put(rule, rule);
}
/*
if (((UnaryRule)rule).child == targetNumberer.number("PRP") &&
((String)sourceNumberer.object(rule.parent)).charAt(0) == 'N') {
System.out.println("Source UR: "+sourceRule+" %% "+sourceRule.score);
System.out.println("Score of "+rule+"is now: "+((UnaryRule)unaryRules.get(rule)).score);
}
*/
}
}
for (UnaryRule ur : unaryRules.keySet()) {
targetUG.addRule(ur);
// System.out.println("UR: "+targetNumberer.object(ur.parent)+" ->
// "+targetNumberer.object(ur.child)+" %% "+ur.score);
}
targetUG.purgeRules();
System.out.println(
"Projected "
+ sourceNumberer.total()
+ " states to "
+ targetNumberer.total()
+ " states.");
}
/**
* traceTo() values that are contained in the tree are assigned to new objects. traceTo() values
* that are not contained in the tree are given the old value.
*/
public static void fixEmptyTreeLeafs(Tree t, Map<Tree, Tree> newToOld, Map<Tree, Tree> oldToNew) {
Tree[] kids = t.children();
for (int i = 0, n = kids.length; i < n; i++) {
fixEmptyTreeLeafs(kids[i], newToOld, oldToNew);
}
if (t instanceof EmptyTreeLeaf) {
EmptyTreeLeaf oldT = (EmptyTreeLeaf) newToOld.get(t);
((EmptyTreeLeaf) t).setEmptyType(oldT.emptyType());
Tree oldTraceTo = oldT.traceTo();
Tree newTraceTo = oldToNew.get(oldTraceTo);
if (newTraceTo != null) ((EmptyTreeLeaf) t).setTraceTo(newTraceTo);
else ((EmptyTreeLeaf) t).setTraceTo(oldTraceTo);
}
}
public static final String doCorefResolution(Annotation annotation) {
Map<Integer, CorefChain> corefs = annotation.get(CorefChainAnnotation.class);
List<CoreMap> sentences = annotation.get(CoreAnnotations.SentencesAnnotation.class);
List<String> resolved = new ArrayList<String>();
for (CoreMap sentence : sentences) {
List<CoreLabel> tokens = sentence.get(CoreAnnotations.TokensAnnotation.class);
for (CoreLabel token : tokens) {
Integer corefClustId = token.get(CorefCoreAnnotations.CorefClusterIdAnnotation.class);
CorefChain chain = corefs.get(corefClustId);
if (chain == null) resolved.add(token.word());
else {
int sentINdx = chain.getRepresentativeMention().sentNum - 1;
CoreMap corefSentence = sentences.get(sentINdx);
List<CoreLabel> corefSentenceTokens = corefSentence.get(TokensAnnotation.class);
CorefMention reprMent = chain.getRepresentativeMention();
if (token.index() < reprMent.startIndex || token.index() > reprMent.endIndex) {
for (int i = reprMent.startIndex; i < reprMent.endIndex; i++) {
CoreLabel matchedLabel = corefSentenceTokens.get(i - 1);
resolved.add(matchedLabel.word());
}
} else resolved.add(token.word());
}
}
}
String resolvedStr = "";
System.out.println();
for (String str : resolved) {
resolvedStr += str + " ";
}
System.out.println(resolvedStr);
return resolvedStr;
}
public Object intern(Object o) {
Object i = oToO.get(o);
if (i == null) {
i = o;
oToO.put(o, o);
}
return i;
}
protected String historyToString(List history) {
String str = (String) historyToString.get(history);
if (str == null) {
StringBuilder sb = new StringBuilder();
for (int i = 0; i < history.size(); i++) {
sb.append('^');
sb.append(history.get(i));
}
str = sb.toString();
historyToString.put(history, str);
}
return str;
}
public Object formResult() {
Set brs = new HashSet();
Set urs = new HashSet();
// scan each rule / history pair
int ruleCount = 0;
for (Iterator pairI = rulePairs.keySet().iterator(); pairI.hasNext(); ) {
if (ruleCount % 100 == 0) {
System.err.println("Rules multiplied: " + ruleCount);
}
ruleCount++;
Pair rulePair = (Pair) pairI.next();
Rule baseRule = (Rule) rulePair.first;
String baseLabel = (String) ruleToLabel.get(baseRule);
List history = (List) rulePair.second;
double totalProb = 0;
for (int depth = 1; depth <= HISTORY_DEPTH() && depth <= history.size(); depth++) {
List subHistory = history.subList(0, depth);
double c_label = labelPairs.getCount(new Pair(baseLabel, subHistory));
double c_rule = rulePairs.getCount(new Pair(baseRule, subHistory));
// System.out.println("Multiplying out "+baseRule+" with history "+subHistory);
// System.out.println("Count of "+baseLabel+" with "+subHistory+" is "+c_label);
// System.out.println("Count of "+baseRule+" with "+subHistory+" is "+c_rule );
double prob = (1.0 / HISTORY_DEPTH()) * (c_rule) / (c_label);
totalProb += prob;
for (int childDepth = 0; childDepth <= Math.min(HISTORY_DEPTH() - 1, depth); childDepth++) {
Rule rule = specifyRule(baseRule, subHistory, childDepth);
rule.score = (float) Math.log(totalProb);
// System.out.println("Created "+rule+" with score "+rule.score);
if (rule instanceof UnaryRule) {
urs.add(rule);
} else {
brs.add(rule);
}
}
}
}
System.out.println("Total states: " + stateNumberer.total());
BinaryGrammar bg = new BinaryGrammar(stateNumberer.total());
UnaryGrammar ug = new UnaryGrammar(stateNumberer.total());
for (Iterator brI = brs.iterator(); brI.hasNext(); ) {
BinaryRule br = (BinaryRule) brI.next();
bg.addRule(br);
}
for (Iterator urI = urs.iterator(); urI.hasNext(); ) {
UnaryRule ur = (UnaryRule) urI.next();
ug.addRule(ur);
}
return new Pair(ug, bg);
}
/**
* Prints out all matches of a semgrex pattern on a file of dependencies. <br>
* Usage:<br>
* java edu.stanford.nlp.semgraph.semgrex.SemgrexPattern [args] <br>
* See the help() function for a list of possible arguments to provide.
*/
public static void main(String[] args) throws IOException {
Map<String, Integer> flagMap = Generics.newHashMap();
flagMap.put(PATTERN, 1);
flagMap.put(TREE_FILE, 1);
flagMap.put(MODE, 1);
flagMap.put(EXTRAS, 1);
flagMap.put(CONLLU_FILE, 1);
flagMap.put(OUTPUT_FORMAT_OPTION, 1);
Map<String, String[]> argsMap = StringUtils.argsToMap(args, flagMap);
args = argsMap.get(null);
// TODO: allow patterns to be extracted from a file
if (!(argsMap.containsKey(PATTERN)) || argsMap.get(PATTERN).length == 0) {
help();
System.exit(2);
}
SemgrexPattern semgrex = SemgrexPattern.compile(argsMap.get(PATTERN)[0]);
String modeString = DEFAULT_MODE;
if (argsMap.containsKey(MODE) && argsMap.get(MODE).length > 0) {
modeString = argsMap.get(MODE)[0].toUpperCase();
}
SemanticGraphFactory.Mode mode = SemanticGraphFactory.Mode.valueOf(modeString);
String outputFormatString = DEFAULT_OUTPUT_FORMAT;
if (argsMap.containsKey(OUTPUT_FORMAT_OPTION) && argsMap.get(OUTPUT_FORMAT_OPTION).length > 0) {
outputFormatString = argsMap.get(OUTPUT_FORMAT_OPTION)[0].toUpperCase();
}
OutputFormat outputFormat = OutputFormat.valueOf(outputFormatString);
boolean useExtras = true;
if (argsMap.containsKey(EXTRAS) && argsMap.get(EXTRAS).length > 0) {
useExtras = Boolean.valueOf(argsMap.get(EXTRAS)[0]);
}
List<SemanticGraph> graphs = Generics.newArrayList();
// TODO: allow other sources of graphs, such as dependency files
if (argsMap.containsKey(TREE_FILE) && argsMap.get(TREE_FILE).length > 0) {
for (String treeFile : argsMap.get(TREE_FILE)) {
System.err.println("Loading file " + treeFile);
MemoryTreebank treebank = new MemoryTreebank(new TreeNormalizer());
treebank.loadPath(treeFile);
for (Tree tree : treebank) {
// TODO: allow other languages... this defaults to English
SemanticGraph graph =
SemanticGraphFactory.makeFromTree(
tree,
mode,
useExtras
? GrammaticalStructure.Extras.MAXIMAL
: GrammaticalStructure.Extras.NONE,
true);
graphs.add(graph);
}
}
}
if (argsMap.containsKey(CONLLU_FILE) && argsMap.get(CONLLU_FILE).length > 0) {
CoNLLUDocumentReader reader = new CoNLLUDocumentReader();
for (String conlluFile : argsMap.get(CONLLU_FILE)) {
System.err.println("Loading file " + conlluFile);
Iterator<SemanticGraph> it = reader.getIterator(IOUtils.readerFromString(conlluFile));
while (it.hasNext()) {
SemanticGraph graph = it.next();
graphs.add(graph);
}
}
}
for (SemanticGraph graph : graphs) {
SemgrexMatcher matcher = semgrex.matcher(graph);
if (!(matcher.find())) {
continue;
}
if (outputFormat == OutputFormat.LIST) {
System.err.println("Matched graph:");
System.err.println(graph.toString(SemanticGraph.OutputFormat.LIST));
boolean found = true;
while (found) {
System.err.println(
"Matches at: " + matcher.getMatch().value() + "-" + matcher.getMatch().index());
List<String> nodeNames = Generics.newArrayList();
nodeNames.addAll(matcher.getNodeNames());
Collections.sort(nodeNames);
for (String name : nodeNames) {
System.err.println(
" "
+ name
+ ": "
+ matcher.getNode(name).value()
+ "-"
+ matcher.getNode(name).index());
}
System.err.println();
found = matcher.find();
}
} else if (outputFormat == OutputFormat.OFFSET) {
if (graph.vertexListSorted().isEmpty()) {
continue;
}
System.out.printf(
"+%d %s%n",
graph.vertexListSorted().get(0).get(CoreAnnotations.LineNumberAnnotation.class),
argsMap.get(CONLLU_FILE)[0]);
}
}
}
/**
* Do max language model markov segmentation. Note that this algorithm inherently tags words as it
* goes, but that we throw away the tags in the final result so that the segmented words are
* untagged. (Note: for a couple of years till Aug 2007, a tagged result was returned, but this
* messed up the parser, because it could use no tagging but the given tagging, which often wasn't
* very good. Or in particular it was a subcategorized tagging which never worked with the current
* forceTags option which assumes that gold taggings are inherently basic taggings.)
*
* @param s A String to segment
* @return The list of segmented words.
*/
private ArrayList<HasWord> segmentWordsWithMarkov(String s) {
int length = s.length();
// Set<String> POSes = (Set<String>) POSDistribution.keySet(); // 1.5
int numTags = POSes.size();
// score of span with initial word of this tag
double[][][] scores = new double[length][length + 1][numTags];
// best (length of) first word for this span with this tag
int[][][] splitBacktrace = new int[length][length + 1][numTags];
// best tag for second word over this span, if first is this tag
int[][][] POSbacktrace = new int[length][length + 1][numTags];
for (int i = 0; i < length; i++) {
for (int j = 0; j < length + 1; j++) {
Arrays.fill(scores[i][j], Double.NEGATIVE_INFINITY);
}
}
// first fill in word probabilities
for (int diff = 1; diff <= 10; diff++) {
for (int start = 0; start + diff <= length; start++) {
int end = start + diff;
StringBuilder wordBuf = new StringBuilder();
for (int pos = start; pos < end; pos++) {
wordBuf.append(s.charAt(pos));
}
String word = wordBuf.toString();
for (String tag : POSes) {
IntTaggedWord itw = new IntTaggedWord(word, tag, wordIndex, tagIndex);
double score = lex.score(itw, 0, word, null);
if (start == 0) {
score += Math.log(initialPOSDist.probabilityOf(tag));
}
scores[start][end][itw.tag()] = score;
splitBacktrace[start][end][itw.tag()] = end;
}
}
}
// now fill in word combination probabilities
for (int diff = 2; diff <= length; diff++) {
for (int start = 0; start + diff <= length; start++) {
int end = start + diff;
for (int split = start + 1; split < end && split - start <= 10; split++) {
for (String tag : POSes) {
int tagNum = tagIndex.indexOf(tag, true);
if (splitBacktrace[start][split][tagNum] != split) {
continue;
}
Distribution<String> rTagDist = markovPOSDists.get(tag);
if (rTagDist == null) {
continue; // this happens with "*" POS
}
for (String rTag : POSes) {
int rTagNum = tagIndex.indexOf(rTag, true);
double newScore =
scores[start][split][tagNum]
+ scores[split][end][rTagNum]
+ Math.log(rTagDist.probabilityOf(rTag));
if (newScore > scores[start][end][tagNum]) {
scores[start][end][tagNum] = newScore;
splitBacktrace[start][end][tagNum] = split;
POSbacktrace[start][end][tagNum] = rTagNum;
}
}
}
}
}
}
int nextPOS = ArrayMath.argmax(scores[0][length]);
ArrayList<HasWord> words = new ArrayList<HasWord>();
int start = 0;
while (start < length) {
int split = splitBacktrace[start][length][nextPOS];
StringBuilder wordBuf = new StringBuilder();
for (int i = start; i < split; i++) {
wordBuf.append(s.charAt(i));
}
String word = wordBuf.toString();
// String tag = tagIndex.get(nextPOS);
// words.add(new TaggedWord(word, tag));
words.add(new Word(word));
if (split < length) {
nextPOS = POSbacktrace[start][length][nextPOS];
}
start = split;
}
return words;
}