Towards Conversational Recommendation over Multi-Type Dialogs

TL;DR

This paper introduces a new task of conversational recommendation over multi-type dialogs, supported by a Chinese dataset, and establishes baseline results to facilitate future research in natural, proactive recommendation dialogues.

Contribution

It presents a novel task of conversational recommendation in multi-type dialogs and provides a large, annotated dataset for systematic study and baseline evaluations.

Findings

Created DuRecDial dataset with 10k dialogs and 156k utterances

Demonstrated baseline performance on the new task

Facilitated future research in proactive conversational recommendation

Abstract

We propose a new task of conversational recommendation over multi-type dialogs, where the bots can proactively and naturally lead a conversation from a non-recommendation dialog (e.g., QA) to a recommendation dialog, taking into account user's interests and feedback. To facilitate the study of this task, we create a human-to-human Chinese dialog dataset \emph{DuRecDial} (about 10k dialogs, 156k utterances), which contains multiple sequential dialogs for every pair of a recommendation seeker (user) and a recommender (bot). In each dialog, the recommender proactively leads a multi-type dialog to approach recommendation targets and then makes multiple recommendations with rich interaction behavior. This dataset allows us to systematically investigate different parts of the overall problem, e.g., how to naturally lead a dialog, how to interact with users for recommendation. Finally we establish baseline results on DuRecDial for future studies. Dataset and codes are publicly available at https://github.com/PaddlePaddle/models/tree/develop/PaddleNLP/Research/ACL2020-DuRecDial.