Cross-Modal Information Retrieval

Cross-Modal Information Retrieval (CMIR) refers to the task of finding relevant items across different data types. Its core objective is to learn a latent representation space to bridge the heterogeneity gap, thereby enabling direct measurement of similarity between items from different modalities. CMIR has significant application value in multimedia information retrieval, recommendation systems, and human-computer interaction.