The disclosed systems and methods provide a novel framework that parameterizes columns in a templated virtual semantic layer. The disclosed framework enables inter- and cross-column relationships between stored data within a SQL database to be determined and stored, and then leveraged at query time to enable an optimized search of the data within the database. The disclosed framework enables a database to be construed, modified and/or configured according to a hierarchy, and/or with types of metadata, that corresponds to the determined column relationships. This not only reduces the amount of data stored and reduces the time to process queries, but also enables a more streamlined approach for searches to be constructed and executed, which can increase the performance of the database and the operations of systems that are backed by the database.