protected void visitCell(final HashSetChartCell cell) {
final int start = cell.start(), end = cell.end();
Collection<Production> possibleProds;
ChartEdge edge;
final ChartEdge[] bestEdges = new ChartEdge[grammar.numNonTerms()]; // inits to null
// final int maxEdgesToAdd = (int) opts.param2;
final int maxEdgesToAdd = Integer.MAX_VALUE;
for (int mid = start + 1; mid <= end - 1; mid++) { // mid point
final HashSetChartCell leftCell = chart.getCell(start, mid);
final HashSetChartCell rightCell = chart.getCell(mid, end);
for (final int leftNT : leftCell.getLeftChildNTs()) {
for (final int rightNT : rightCell.getRightChildNTs()) {
possibleProds = grammar.getBinaryProductionsWithChildren(leftNT, rightNT);
if (possibleProds != null) {
for (final Production p : possibleProds) {
edge = chart.new ChartEdge(p, leftCell, rightCell);
addEdgeToArray(edge, bestEdges);
}
}
}
}
}
addBestEdgesToChart(cell, bestEdges, maxEdgesToAdd);
}
public void addUnaryExtensionsToLexProds() {
for (int i = 0; i < chart.size(); i++) {
final HashSetChartCell cell = chart.getCell(i, i + 1);
for (final int pos : cell.getPosNTs()) {
for (final Production unaryProd : grammar.getUnaryProductionsWithChild(pos)) {
// cell.addEdge(unaryProd, cell, null, cell.getBestEdge(pos).inside + unaryProd.prob);
cell.updateInside(unaryProd, cell.getInside(pos) + unaryProd.prob);
}
}
}
}
protected void initParser(final ParseTask parseTask) {
chart = new CellChart(parseTask, this);
this.maxEdgeFOM = new float[chart.size()][chart.size() + 1];
this.spanAgenda = new PriorityQueue<HashSetChartCell>();
// The chart is (chart.size()+1)*chart.size()/2
for (int start = 0; start < chart.size(); start++) {
for (int end = start + 1; end < chart.size() + 1; end++) {
maxEdgeFOM[start][end] = Float.NEGATIVE_INFINITY;
}
}
}
protected void expandFrontier(final HashSetChartCell cell) {
// connect edge as possible right non-term
for (int start = 0; start < cell.start(); start++) {
setSpanMaxEdgeFOM(chart.getCell(start, cell.start()), cell);
}
// connect edge as possible left non-term
for (int end = cell.end() + 1; end <= chart.size(); end++) {
setSpanMaxEdgeFOM(cell, chart.getCell(cell.end(), end));
}
}
protected void addLexicalProductions(final int sent[]) {
// ChartEdge newEdge;
HashSetChartCell cell;
for (int i = 0; i < chart.size(); i++) {
cell = chart.getCell(i, i + 1);
for (final Production lexProd : grammar.getLexicalProductionsWithChild(sent[i])) {
// newEdge = chart.new ChartEdge(lexProd, chart.getCell(i, i + 1));
// chart.getCell(i, i + 1).addEdge(newEdge);
cell.updateInside(lexProd, lexProd.prob);
}
}
}
protected HashSetChartCell next() {
// return spanAgenda.poll();
HashSetChartCell bestSpan = null;
float bestScore = Float.NEGATIVE_INFINITY;
for (int span = 1; span <= chart.size(); span++) {
for (int beg = 0; beg < chart.size() - span + 1; beg++) { // beginning
if (maxEdgeFOM[beg][beg + span] > bestScore) {
bestScore = maxEdgeFOM[beg][beg + span];
bestSpan = chart.getCell(beg, beg + span);
}
}
}
return bestSpan;
}
@Override
public BinaryTree<String> findBestParse(final ParseTask parseTask) {
HashSetChartCell cell;
initParser(parseTask);
addLexicalProductions(parseTask.tokens);
figureOfMerit.initSentence(parseTask, chart);
addUnaryExtensionsToLexProds();
for (int i = 0; i < chart.size(); i++) {
expandFrontier(chart.getCell(i, i + 1));
}
while (hasNext() && !chart.hasCompleteParse(grammar.startSymbol)) {
cell = next();
// System.out.println(" nextCell: " + cell);
visitCell(cell);
expandFrontier(cell);
}
return chart.extractBestParse(grammar.startSymbol);
}
protected void setSpanMaxEdgeFOM(
final HashSetChartCell leftCell, final HashSetChartCell rightCell) {
ChartEdge edge;
final int start = leftCell.start(), end = rightCell.end();
float bestFOM = maxEdgeFOM[start][end];
// System.out.println(" setSpanMax: " + leftCell + " && " + rightCell);
Collection<Production> possibleProds;
for (final int leftNT : leftCell.getLeftChildNTs()) {
for (final int rightNT : rightCell.getRightChildNTs()) {
possibleProds = grammar.getBinaryProductionsWithChildren(leftNT, rightNT);
if (possibleProds != null) {
for (final Production p : possibleProds) {
// final float prob = p.prob + leftCell.getInside(leftNT) +
// rightCell.getInside(rightNT);
edge = chart.new ChartEdge(p, leftCell, rightCell);
// System.out.println(" considering: " + edge);
if (edge.fom > bestFOM) {
bestFOM = edge.fom;
}
}
}
}
}
if (bestFOM > maxEdgeFOM[start][end]) {
final HashSetChartCell parentCell = chart.getCell(start, end);
// if (maxEdgeFOM[start][end] > Float.NEGATIVE_INFINITY) {
// spanAgenda.remove(parentCell);
// }
maxEdgeFOM[start][end] = bestFOM;
parentCell.fom = bestFOM;
// spanAgenda.add(parentCell);
// System.out.println(" addingSpan: " + parentCell);
}
}
@Override
public float getInside(final int start, final int end, final int nt) {
return chart.getInside(start, end, nt);
}