Exploring Speaker-Related Information in Spoken Language Understanding for Better Speaker Diarization