/** Process discourse information */
protected void processDiscourse(Dictionaries dict) {
docType = findDocType(dict);
markQuotations(this.annotation.get(CoreAnnotations.SentencesAnnotation.class), false);
findSpeakers(dict);
// find 'speaker mention' for each mention
for (Mention m : allPredictedMentions.values()) {
int utter = m.headWord.get(CoreAnnotations.UtteranceAnnotation.class);
String speaker = m.headWord.get(CoreAnnotations.SpeakerAnnotation.class);
if (speaker != null) {
// Populate speaker info
SpeakerInfo speakerInfo = speakerInfoMap.get(speaker);
if (speakerInfo == null) {
speakerInfoMap.put(speaker, speakerInfo = new SpeakerInfo(speaker));
// span indicates this is the speaker
if (Rules.mentionMatchesSpeaker(m, speakerInfo, true)) {
m.speakerInfo = speakerInfo;
}
}
if (NumberMatchingRegex.isDecimalInteger(speaker)) {
try {
int speakerMentionID = Integer.parseInt(speaker);
if (utter != 0) {
// Add pairs of mention id and the mention id of the speaker
speakerPairs.add(new Pair<Integer, Integer>(m.mentionID, speakerMentionID));
// speakerPairs.add(new Pair<Integer, Integer>(speakerMentionID,
// m.mentionID));
}
} catch (Exception e) {
// no mention found for the speaker
// nothing to do
}
}
}
// set generic 'you' : e.g., you know in conversation
if (docType != DocType.ARTICLE
&& m.person == Person.YOU
&& m.endIndex < m.sentenceWords.size() - 1
&& m.sentenceWords
.get(m.endIndex)
.get(CoreAnnotations.TextAnnotation.class)
.equalsIgnoreCase("know")) {
m.generic = true;
}
}
// now that we have identified the speakers, first pass to check if mentions should cluster with
// the speakers
for (Mention m : allPredictedMentions.values()) {
if (m.speakerInfo == null) {
for (SpeakerInfo speakerInfo : speakerInfoMap.values()) {
if (speakerInfo.hasRealSpeakerName()) {
// do loose match - assumes that there isn't that many speakers....
if (Rules.mentionMatchesSpeaker(m, speakerInfo, false)) {
m.speakerInfo = speakerInfo;
break;
}
}
}
}
}
}
