Abstract: This study aims to construct a multimodal knowledge graph in the field of reading nourishment, representing domain knowledge as a network of entities, attributes, and relationships to enable crossmodal integration and visualization of reading nourishment books. The knowledge graph supports access and indepth querying, thereby providing a knowledge foundation for reading promotion aimed at psychological development. This paper introduces ontologies to build a semantic description model for reading nourishment books oriented towards psychological development. First, data acquisition and preprocessing are conducted. The data sources include empirically validated, publicly recommended, and expertrecommended books, music, and audiovisual materials that are beneficial for psychological development. Second, knowledge extraction and mining techniques are used to reorganize knowledge from relevant texts, videos, and audio resources. Finally, based on the constructed knowledge graph, the paper explores the implementation pathways for large models in the reading nourishment field. The semantic association scheme for the reading nourishment domain for multimodal books is feasible and applicable, promoting multidimensional knowledge organization and fusion applications across scenarios, resources, and spatiotemporal contexts in the field of psychological development. The knowledge graph construted in this study provides a knowledge base for AIbased dialogue repositories in the reading nourishment domain and supplies highquality external knowledge for domainspecific large language models, thereby better serving the national strategy of “deepening national reading” and the societal needs for psychological development.