Table 1.

Example dialogues with their respective informativeness (I) and linguistic quality (Q) scores.

Example question and answer pair (path in word graph)a	Informativeness score (I)	Linguistic quality score (Q)	Included in summary
Participant was asked have they been diagnosed with depression participant said yeah while ago	0.25	0.1	Yes
Participant was asked uh huh, then participant said pretty easy	0.08	0.04	No

^aSentences containing words that are semantically close to those in the depression lexicon have higher Q than those that do not. The last column says whether the integer linear programming framework selected the sentence or not based on I and Q.