Automatic Data Model Generation from SQL Statements

The system performs preliminary actions by executing the SQL query against the database to retrieve actual data results before generating the data model. This preliminary data retrieval enables the system to analyze implicit information in the data itself, such as inferred relationships and patterns, rather than relying solely on explicit query definitions. The retrieved data serves as a foundation for discovering additional semantic information that would otherwise remain hidden.

The patent introduces an intermediary data analysis layer between the SQL query and the final data model. This intermediary processes the actual query results to extract implicit information, such as inferred relationships between entities, cardinality constraints, and semantic meanings that are not explicitly defined in the query structure. This intermediary analysis bridge connects the explicit query syntax with the implicit data semantics.

The system extracts only the essential and relevant implicit information from the query results rather than processing all possible data attributes. It identifies and extracts key relationships, entity types, and cardinality constraints that are most important for building an accurate data model, while ignoring redundant or less significant information. This selective extraction reduces processing overhead while maintaining model quality.

The patent applies partial action by focusing the analysis on specific aspects of the query results that are most indicative of the database structure, such as primary key relationships, foreign key constraints, and entity-cardinality patterns. Rather than exhaustively analyzing every possible data attribute and relationship, the system concentrates computational effort on the most informative subsets of the data, achieving good model quality with reduced processing time.

The system segments the comprehensive set of fields into logical groups based on their functional relationships and semantic meanings. It organizes fields into entity types, relationship types, and attribute categories, making the large number of fields more manageable and easier to navigate. This segmentation transforms a flat, overwhelming list of fields into a structured, hierarchical organization that improves usability.

The patent adds another dimension to field organization by introducing semantic categorization and relationship-based grouping beyond simple table membership. Fields are organized not just by their source table but also by their role in relationships, their semantic category, and their connectivity to other fields. This multi-dimensional organization makes it easier to manage and understand large numbers of fields by providing multiple ways to navigate and filter them.