Pearl: A Review-driven Persona-Knowledge Grounded Conversational Recommendation Dataset

TL;DR

PEARL is a large-scale conversational recommendation dataset that incorporates detailed user personas and knowledge, enabling more specific, expert, and contextually relevant recommendations in dialogue systems.

Contribution

The paper introduces PEARL, a novel dataset synthesized with persona- and knowledge-augmented LLMs, addressing gaps in existing datasets by including user preferences and explanations.

Findings

PEARL dialogues contain more specific user preferences.

PEARL demonstrates higher domain expertise in utterances.

PEARL yields more relevant recommendations than prior datasets.

Abstract

Conversational recommender system is an emerging area that has garnered an increasing interest in the community, especially with the advancements in large language models (LLMs) that enable diverse reasoning over conversational input. Despite the progress, the field has many aspects left to explore. The currently available public datasets for conversational recommendation lack specific user preferences and explanations for recommendations, hindering high-quality recommendations. To address such challenges, we present a novel conversational recommendation dataset named PEARL, synthesized with persona- and knowledge-augmented LLM simulators. We obtain detailed persona and knowledge from real-world reviews and construct a large-scale dataset with over 57k dialogues. Our experimental results demonstrate that utterances in PEARL include more specific user preferences, show expertise in the target domain, and provide recommendations more relevant to the dialogue context than those in prior datasets.