In this dataset, image-text pairs of collection data were crawled from the official website of the British Museum with the keyword "ceramic" in China, and the English text was translated and refined. The wrong expression of vessel names, patterns and crafts was corrected, and the word order was adjusted to form the structure of "type - texture pattern - craft".The hand-sifted dataset contains a total of 4,096 image-text pairs, covering a variety of ceramic categories from the Stone Age to the recent past, such as teacups, dishes, statues, vases, and more.