Online dating services have become popular in modern society. Pair matching prediction between two users in these services can help efficiently increase the probability of finding their life partners. Deep learning based methods with automatic feature interaction functions such as Factorization Machines (FMs) and cross network of Deep & Cross Network (DCN) can model sparse categorical features, which are effective to many prediction tasks of web applications. To solve the pair matching task, we improve these FM-based and DCN-based deep models by enhancing the representation of feature interaction embedding and proposing a novel design of interaction pooling layer avoiding information loss. Through the experiments on a real service dataset, we demonstrate the superior performances of our proposed models.