Multimodal Link Prediction Method for Commodity Knowledge Graph

Litao Zhang, Chunming Yang, Hui Zhang, Honghui Xiang

Abstract: Knowledge graph link prediction is to predict whether there is a relation between two entities according to the existing knowledge graph data, which is of great significance for improving the existing knowledge graph. Based on the application of knowledge graph in electronic commerce, this paper proposes a new multimodal link prediction method for commodity knowledge graph. This method uses the visual features extracted from commodity images by the pre-trained visual model, then integrates entity embedding to enrich entity semantics, and finally learns entity and relation representation by knowledge graph embedding method. The experimental results show that the proposed method achieves better results than the existing methods in the task of commodity knowledge graph link prediction, which proves the effectiveness of the multimodal fusion method in the knowledge graph link prediction.