Social networking services (SNS) contain vast amounts of image-text posts, necessitating effective analysis of their relationships for improved information retrieval. This study addresses the classification of image-text pairs in SNS, overcoming prior limitations in distinguishing relationships beyond similarity. We propose a graph-based method to classify image-text pairs into similar and complementary relationships. Our approach first embeds images and text using CLIP, followed by clustering. Next, we construct an Image-Text Relationship Clustering Line Graph (ITRC-Line Graph), where clusters serve as nodes. Finally, edges and nodes are swapped in a pseudo-graph representation. A Graph Convolutional Network (GCN) then learns node and edge representations, which are fused with the original embeddings for final classification. Experimental results on a publicly available dataset demonstrate the effectiveness of our method.