In the whole brain architecture approach toward AGI, building a machine learning algorithm that corresponds to local circuit of the hippocampus is indispensable. However, extreme complexity of the hippocampus stands in the way. In this article, as a preliminary step toward that, we build a framework of the entorhinal cortex(EC), the gateway of information flow of the hippocampus. Based on neuroscientific knowledge, we define the interface and semantics of input to and output from the EC and specify the functions of the EC. Then, with this framework, we discuss the plausibility of existing computational hypotheses.