public void process(Context c, Sentence s) {
// we assume markers occur only at the beginning of clauses
// TODO is this heuristic language-specific?
Intervals markerInts = new Intervals();
for (Interval clause : s.get(Features.CLAUSES))
for (Marker m : markers)
if (testMarker(s, clause, m)) {
markerInts.add(clause.low(), clause.low() + m.length()).set(Features.MARKER_TYPE, m.type);
break;
}
s.set(Features.DISCOURSE_MARKERS, markerInts);
}
private void split(Sentence s, Interval clause, Intervals out) {
Interval pre = null, verb = null, post = null;
for (Interval phrase : s.get(Features.PHRASES).in(clause)) {
if (verb == null) {
if (phrase.get(Features.PHRASE_TYPE) == PhraseType.VerbPhrase)
verb = new Interval(phrase.low(), phrase.high());
else pre = expand(pre, phrase);
} else post = expand(post, phrase);
}
boolean preFocus = anyFocused(s, pre);
boolean verbFocus = anyFocused(s, verb);
boolean postFocus = anyFocused(s, post);
// heuristics from Hiyakumoto, Prevost and Cassell
if (!preFocus && !verbFocus && !postFocus) { // case 1
maybeAdd(out, Theme, pre);
maybeAdd(out, Rheme, verb, post);
} else if (preFocus && verbFocus && !postFocus) { // case 2
maybeAdd(out, Theme, pre);
maybeAdd(out, Rheme, verb, post);
} else if (preFocus && !verbFocus && postFocus) { // case 3
maybeAdd(out, Theme, pre, verb);
maybeAdd(out, Rheme, post);
} else if (!preFocus && verbFocus && postFocus) { // case 4
maybeAdd(out, Theme, pre, verb);
maybeAdd(out, Rheme, post);
} else if (preFocus && verbFocus && postFocus) { // case 5
maybeAdd(out, Theme, pre, verb);
maybeAdd(out, Rheme, post);
} else if (preFocus && !verbFocus && !postFocus) { // case 6
maybeAdd(out, Rheme, pre);
maybeAdd(out, Theme, verb, post);
} else if (!preFocus && verbFocus && !postFocus) { // case 7
maybeAdd(out, Theme, pre);
maybeAdd(out, Rheme, verb, post);
} else if (!preFocus && !verbFocus && postFocus) { // case 8
maybeAdd(out, Theme, pre, verb);
maybeAdd(out, Rheme, post);
}
}
private Interval expand(Interval x, Interval y) {
if (x == null) return y;
else if (y == null) return x;
else return new Interval(Math.min(x.low(), y.low()), Math.max(x.high(), y.high()));
}
