/**
* This method adds a new Mention and associated concepts to a SofaText, there can be overlap
* between concept names and multiple concepts can match a string, instead of choosing one
* concept, we record them all.
*
* @param m
* @param concepts
*/
public void addMentionAndConcepts(Mention m, List<Concept> concepts) {
// System.out.println("Mention Text: " + m.getText());
ArrayList<SofaTextMention> toRemove = new ArrayList<SofaTextMention>();
// check all SofaTextMentions
for (SofaTextMention stm : sofaTextMention) {
// if new mention is subsumed by existing mention, add associated concepts to subsuming
// mention
// and then return
// test for overlap
if (isOverlapping(stm, m)) { // there is overlap, keep the larger of the two mentions
if (stm.getMentionEnd() - stm.getMentionStart()
< m.getEnd() - m.getStart()) // mention is larger
{
System.out.println("1");
concepts.addAll(stm.getConcepts());
toRemove.add(stm);
} else {
System.out.println("2");
stm.addConcepts(concepts);
return;
}
}
}
// remove mentions from remove list
sofaTextMention.removeAll(toRemove);
// add new mention to sofatext
sofaTextMention.add(new SofaTextMention(this, m, concepts));
}
private boolean isOverlapping(SofaTextMention stm, Mention m) {
if (stm.getMentionStart() == m.getStart() // same start token
|| stm.getMentionEnd() == m.getEnd() // same end token
|| (stm.getMentionStart() <= m.getStart() && stm.getMentionEnd() >= m.getStart())
|| (stm.getMentionStart() >= m.getStart() && stm.getMentionStart() <= m.getEnd())) {
return true;
}
return false;
}