[Purpose/significance] In the Internet landscape, multimodal User-Generated Content (UGC) comprising images and text on mobile social media has emerged as a crucial information source influencing users' decision-mak⁃ ing. However, existing studies on user information adoption have analyzed multimodal content such as images and text in a piecemeal fashion, leading to a lack of clarity regarding the comprehensive influence mechanism of multidimen⁃ sional UGC features, especially fusion features, on user information adoption. [Method/process] This study employs Xiaohongshu, a representative experience-sharing mobile social media platform, as the research object. We further im⁃ prove the information analysis method by utilizing NLP and CV technologies. Specifically, we propose a multidimen⁃ sional feature measurement method for multimodal UGC, which captures both images and text's deep, shallow, and fu⁃ sion features. Furthermore, we develop an econometric model to investigate the influence mechanism of multidimen⁃ sional features on user information adoption. [Result/conclusion] The results indicate that the impact of two image-tex⁃ tharmony features on user information adoption is significantly higher than that of image or text features in isolation.Ad⁃ ditionally, the influence of shallow features is greater than that of deep features; the key features affecting user informa⁃ tion adoption exhibit moderating effects; and the specific influence mechanisms differ in the three types of scenarios: nature, object, and others. The findings of this study provide crucial insights for facilitating high-quality UGC creation and effective information adoption in mobile social media.