With the rapid development of mobile applications, effectively discovering associations between mobile applications has become an important issue. Existing methods for discovering associations among mobile applications are limited to associations within the same application store. Besides, the methods for association discovery are restricted to a single type of representation learning approach, resulting in suboptimal results. To address this problem, this paper explores two mobile application association discovery frameworks based on representation learning for multi-source mobile applications, including iterative framework combining knowledge graph representation methods and network representation models, and entity alignment-based models to mine associations between mobile apps for this task. Experiments indicate that the proposed methods can obtain better performances than existing methods in terms of metrics.