Built independently by an author, for readers. Read the story and support ChapterPal
ChapterPal Logo
Log inSign up

keyword

PAED dataset

The PAED dataset, or Persona Attribute Extraction in Dialogues dataset, is a natural language processing benchmark designed for identifying and extracting structured personal attributes from conversational text. Developed to advance personalized human-computer interaction and conversational artificial intelligence, the dataset annotates user characteristics, preferences, and traits in the form of structured subject-relation-object triplets derived from dialogue utterances. It provides standardized, high-quality annotations that resolve ambiguities and inconsistencies found in earlier conversational persona corpora by applying precise text-to-label matching criteria. The dataset is primarily utilized to train, evaluate, and benchmark information extraction models, especially in zero-shot learning scenarios where systems must detect previously unseen persona attributes across multi-turn dialogues.

1 item

PAED: Zero-Shot Persona Attribute Extraction in Dialogues

PAED: Zero-Shot Persona Attribute Extraction in Dialogues

Luyao Zhu, Wei Li, Rui Mao, Vlad Pandelea, Erik Cambria

OrganizationsNanyang Technological University

Why you should read this

Introduces a refined persona attribute triplet dataset alongside a generation-based framework that uses a variational autoencoder hard negative sampling strategy to extract structured user personas in zero-shot dialogue settings.

Persona attribute extraction is critical for personalized human-computer interaction. Dialogue is an important medium that communicates and delivers persona information. Although there is a public dataset for triplet-based persona attribute extraction from conversations, its automatically generated labels present many issues, including unspecific relations and inconsistent annotations. We fix such issues by leveraging more reliable text-label matching criteria to generate high-quality data for persona attribute extraction. We also propose a contrastive learning- and generation-based model with a novel hard negative sampling strategy for generalized zero-shot persona attribute extraction. We benchmark our model with state-of-the-art baselines on our dataset and a public dataset, showing outstanding accuracy gains. Our sampling strategy also exceeds others by a large margin in persona attribute extraction.

Added

2026-10-04