Session-based recommendation (SBR) is designed to predict the nextmost likely interactive item based on the current session. The mostadvanced methods in SBR employ graph neural networks and attentionmechanisms to model the adjacent item transition information from othersession sequences and the current session sequence, it employs positionaware attention to incorporate the reversed position information in anitem to learn the importance of each item in the session when generatingthe session representation. But these approaches have some problems.Firstly, uncorrelated items (noise) are inevitably introduced when usinginformation from other sessions. Secondly, reverse position coding makesit difficult to learn the sequence transduction relationship of items inthe session sequence. In this paper, we have proposed a novel sessionbased recommendation method called Transition Information EnhancedGated Graph Neural Network (TIE-GGNN) to solve the above problems. Specifically, TIE-GGNN learns two levels of session embedding,global information enhanced session embedding and transition information enhanced session embedding. The global information enhancedsession representation learning layer (GIESRLL) employs the information of other sessions and the current session to learn global-level sessionembedding, and the transition information enhanced session representation learning layer (TIESRLL) employs the items of the current sessionto learn new session embedding and integrates the time information intothe item representation in the session sequence for neighbor embedding learning, so as to further enhance the sequential transition relationshipin the session sequence. Experiments on three benchmark datasets havedemonstrated that TIE-GGNN is superior to other advanced methods.