With the fast development of wireless communication and sensor technologies, ubiquitous learning can be applied to various subject matters, including art and humanity learning. Among others, learning in museums is an effective way to bring experiential knowledge about works of art to students. For learning in museums, learning sheets are common and useful tools to guide students through the learning process. The learning performance will be improved if learning sheets can be adaptive to different students. However, generating adaptive learning sheets for each student is an overwhelming effort to teachers. This paper focuses on the adaptive learning sheet generation problem for ubiquitous museum guide in grid environments. A knowledge-based architecture is proposed to adaptively generate learning sheets. Also, a grid environment has been built to evaluate the proposed approach.