COMPRISE_Data05_VerbMobil-1Corpus_V1.0

Recordings of two people trying to reach an agreement for the details of a meeting (time, place, activities, etc.) that has been complemented by newly collected privacy-related annotations. Within COMPRISE, we annotated some named entities capturing private information in the conversation like personal name, location, organization, date, and time. We used the annotated corpus to train a named entity recognition model to automatically detect private entities. In total, we annotated 27,000 utterances.