Foundation for Chinese Poetry Research: An Open Large-Scale and Fine-Grained Multimodal Knowledge Graph

ICLR 2026 Conference Submission18467 Authors

Keywords: Multimodal Knowledge Graph, Knowledge Graph Construction, Poetry-Image Retrieval, Classical Chinese poetry, Poetry Question Answering, Poetry Theme Classification

TL;DR: We construct a multimodal knowledge graph on classical Chinese poetry and conduct experiments on three downstream tasks, achieving excellent performance.

Abstract: Classical Chinese poetry is a treasured cultural heritage of humanity, attracting extensive research interest. However, the study of classical Chinese poetry is hindered by the lack of open, large-scale and fine-grained multimodal datasets.Prior datasets are either limited by modality constraints, dataset size, or the level of dataset refinement, making them inadequate for effectively supporting studies and application development of classical Chinese poetry.To address these issues, we propose a method for constructing a large-scale and fine-grained multimodal knowledge graph of classical Chinese poetry. We first design an informative ontology graph for classical Chinese poetry and comprehensively collect poetry knowledge based on it. Furthermore, the method utilizes knowledge augmentation, prompt optimization, and text-image alignment to acquire comprehensive and fine-grained knowledge. Both qualitative and quantitative evaluations are conducted on the Multimodal Knowledge Graph of Classical Chinese Poetry (CPMK), highlighting its comprehensiveness and high quality.We also conduct downstream evaluations on poetry-image retrieval, poetry question answering and poetry theme classification tasks.Significant results were achieved in all three tasks, particularly in poetry-image retrieval and poetry theme classification attained state-of-the-art performance. This outstanding performance highlights the effectiveness of CPMK, which provides a robust foundation for classical Chinese poetry research.CPMK will be released to promote research in Chinese culture.

Primary Area: datasets and benchmarks

Submission Number: 18467