@Override
public void finishTraining() {
lex.finishTraining();
int numTags = tagIndex.size();
POSes = new HashSet<String>(tagIndex.objectsList());
initialPOSDist = Distribution.laplaceSmoothedDistribution(initial, numTags, 0.5);
markovPOSDists = new HashMap<String, Distribution>();
Set entries = ruleCounter.lowestLevelCounterEntrySet();
for (Iterator iter = entries.iterator(); iter.hasNext(); ) {
Map.Entry entry = (Map.Entry) iter.next();
// Map.Entry<List<String>, Counter> entry = (Map.Entry<List<String>, Counter>)
// iter.next();
Distribution d =
Distribution.laplaceSmoothedDistribution((ClassicCounter) entry.getValue(), numTags, 0.5);
markovPOSDists.put(((List<String>) entry.getKey()).get(0), d);
}
}
private Distribution<Integer> getSegmentedWordLengthDistribution(Treebank tb) {
// CharacterLevelTagExtender ext = new CharacterLevelTagExtender();
ClassicCounter<Integer> c = new ClassicCounter<Integer>();
for (Iterator iterator = tb.iterator(); iterator.hasNext(); ) {
Tree gold = (Tree) iterator.next();
StringBuilder goldChars = new StringBuilder();
ArrayList goldYield = gold.yield();
for (Iterator wordIter = goldYield.iterator(); wordIter.hasNext(); ) {
Word word = (Word) wordIter.next();
goldChars.append(word);
}
List<HasWord> ourWords = segment(goldChars.toString());
for (int i = 0; i < ourWords.size(); i++) {
c.incrementCount(Integer.valueOf(ourWords.get(i).word().length()));
}
}
return Distribution.getDistribution(c);
}
/**
* 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;
}
