SIM is a fully featured, commercially available database management system based on a semantic data model similar to Hammer and McLeod's SDM SIM has two primary modeling goals. The first is to narrow the gap between a user's real-world perception of data and the conceptual view imposed by the database system because of modeling presuppositions or limitations. The second goal is to allow, as much as possible, the semantics of data to be defined in the schema and make the database system responsible for enforcing its integrity SIM provides a rich set of constructs for schema definition, including those for specifying generalization hierarchies modeled by directed acyclic graphs, interobject relationships and integrity constraints. It also features a novel, easy-to-use, English-like DML. This paper describes the key modeling features of SIM, the architecture of the system and its implementation considerations.