Answers are grounded in this paper and cite their sources.
This research demonstrates the effectiveness of a model for detecting personal names in call center transcripts, highlighting its importance for data security.
Call centers record and store customer-agent conversations for quality assurance. To ensure data security, it is important to locate and mask sensitive information pertaining to personal details in the transcripts of the audio recordings. In this paper, we describe our development of a model for identifying person names in transcripts from call centers in Hong Kong. We describe the types of names that can potentially appear in the transcript and discuss a careful sampling of negative examples. Our experiment results show that a well fine-tuned transformer-based encoder model can significantly outperform generative large language models.
No takes yet. Share an insight, caveat, or question.
Suen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: