A practical method is presented for eliminating unnecessary database operations that often arise from poorly posed natural language queries. This subsystem consists mainly of a knowledge-base whose rules are semantic constraints of the database. The inference procedure's actions are, unlike that of similar inference engines, strictly controlled by the structure of the query. Because of this it is easy to implement and it is relatively fast. It is believed that this subsystem represents a practical incremental improvement that can be made to any relational database front end, not just those that must cope with natural language queries. The procedure also has value as part of a practical semantic query optimizer.>