An Empirical Study on Zero-Data Bootstrapping for Conversational Recommender Systems

Hugging Face Daily Papers Papers

Summary

This paper presents an empirical study on bootstrapping conversational recommender systems using synthetic data generated from non-conversational signals, demonstrating that it outperforms zero-shot and scarce real-data methods in low-resource settings.

Conversational Recommender Systems (CRS) typically require domain-specific dialogue data, which is costly, scarce, and often unavailable in new domains. We conduct a systematic empirical study of zero-data CRS bootstrapping: generating synthetic conversational supervision from non-conversational signals---item reviews, metadata, and user-item interactions---without any in-domain dialogue corpus. We compare two information-theoretic selection strategies, Jensen-Shannon diversity and Fisher information, across domain signals, model architectures, datasets, and fine-tuning paradigms. Our results show that domain-grounded synthetic data consistently outperforms zero-shot prompting and naive synthetic baselines; active selection improves data efficiency over random sampling; metadata and collaborative filtering signals each improve selection quality; and, in low-resource settings, synthetic data can outperform scarce real dialogues while further complementing them. These findings establish non-conversational domain signals as a viable path toward building CRS without conversational training data. The code is available at https://anonymous.4open.science/r/zero_data_crs/ .
Original Article
View Cached Full Text

Cached at: 09/03/26, 11:58 PM

Paper page - An Empirical Study on Zero-Data Bootstrapping for Conversational Recommender Systems

Source: https://huggingface.co/papers/2504.15476

Abstract

Non-conversational domain signals can generate synthetic dialogue data that outperforms zero-shot and scarce real-data baselines for bootstrapping conversational recommender systems.

Conversational Recommender Systems (CRS) typically require domain-specific dialogue data, which is costly, scarce, and often unavailable in new domains. We conduct a systematic empirical study ofzero-data CRS bootstrapping: generatingsynthetic conversational supervisionfrom non-conversational signals---item reviews, metadata, and user-item interactions---without any in-domain dialogue corpus. We compare two information-theoretic selection strategies,Jensen-Shannon diversityandFisher information, across domain signals, model architectures, datasets, andfine-tuning paradigms. Our results show that domain-grounded synthetic data consistently outperforms zero-shot prompting and naive synthetic baselines;active selectionimproves data efficiency over random sampling; metadata andcollaborative filteringsignals each improve selection quality; and, in low-resource settings, synthetic data can outperform scarce real dialogues while further complementing them. These findings establish non-conversational domain signals as a viable path toward building CRS without conversational training data. The code is available at https://anonymous.4open.science/r/zero_data_crs/ .

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2504\.15476

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2504.15476 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2504.15476 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2504.15476 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

τ-Rec: A Verifiable Benchmark for Agentic Recommender Systems

Hugging Face Daily Papers

τ-Rec is a verifiable benchmark for agentic recommender systems that replaces subjective LLM-as-a-judge evaluations with verifiable rewards and controlled dialogue constraints, revealing steep reliability cliffs across leading models where even the best achieves only ~57% pass@1.