User identity linkage (UIL) aims to link identities belonging to the same individual across various platforms. While numerous methods have been proposed for paired or multiple platforms, UIL is still a non-trivial task due to the following challenges. (1) How to alleviate the sparsity and incompleteness of user information from different platforms? (2) How can UIL approaches achieve high effectiveness while still maintaining low complexity in multi-platform scenarios? In light of these challenges, we propose MMUIL (enhancing multi-platform user identity linkage with multi-information), a novel model excelling in high effectiveness while still maintaining low complexity. The model consists of a Multi-Information Embedding (MIE) module and a Partially Shared Adversarial Learning (PSAL) module. Specifically, for the first challenge, MIE simultaneously considers the token sequence semantics in usernames and the structural information of multi-type networks (i.e., homogeneous and heterogeneous networks). To address the second challenge, the adversarial learning-based PSAL decreases the complexity with shared partial parameters (i.e., shared generators). Meanwhile, to enhance the model’s effectiveness, PSAL exploits an attention mechanism to mitigate the disadvantages of shared partial parameters, such as partial information loss and noise introduction, while integrating the above multi-information intensively. The extensive experiments conducted on two real-world datasets demonstrate that our proposed model MMUIL significantly outperforms the state-of-the-art methods.