QwenSafe: Multimodal Content Rating Description Identification via Preference-Aligned VLMs

TL;DR

QwenSafe is a multimodal vision-language model that automatically identifies content rating descriptors in mobile apps by reasoning over app metadata and screenshots, improving accuracy and consistency.

Contribution

The paper introduces QwenSafe and a data pipeline, metadata2CRD, for scalable training, and demonstrates superior performance over state-of-the-art models in content rating classification.

Findings

QwenSafe outperforms baselines in binary CRD classification.

Model achieves 111.8% improvement in positive-class recall.

Descriptor-aware multimodal alignment enhances content classification accuracy.

Abstract

Mobile app marketplaces require developers to disclose standardized content rating descriptors (CRDs) to inform users about potentially sensitive or restricted content. Ensuring the accuracy and consistency of these disclosures remains challenging due to the multimodal nature of app content, which spans textual descriptions and visual interfaces. In this paper, we present QwenSafe, a Vision-Language Model (VLM) designed to automatically identify the presence of Apple-defined CRDs by jointly reasoning over app metadata and screenshots. To enable scalable training for this task, we introduce metadata2CRD, a data-construction pipeline that synthesizes descriptor-aligned question-answer pairs by combining app descriptions, screenshots, and formal descriptor definitions. We adapt Qwen3-VL-8B using supervised fine-tuning followed by Direct Preference Optimization (DPO) to align model predictions with descriptor-specific evidence and explanations across visual and textual modalities. We evaluate QwenSafe on 12 Apple-defined content rating descriptors and compare it against state-of-the-art vision-language models, including Qwen3-VL, LLaVA-1.6, and Gemini-2.5-Flash. QwenSafe consistently outperforms all baselines in binary CRD classification, achieving improvements in positive-class recall of 111.8%, 36.1%, and 2.1%, respectively. Our results demonstrate that descriptor-aware multimodal alignment substantially improves automated content classification and highlights the potential of vision-language models to support scalable and consistent content rating in mobile app marketplaces.