The schema is the data structure of the Event Store database, which contains column-based tables. Views are created to facilitate efficient data querying with human-readable labels and improved security. Syntasa has a default base schema configured to provide event, session, daily visitor, and product-level analysis. The schema is customizable with the ability for customers to design the structure and template for subsequent apps.
Below describes the tables and views created as part of the pipeline configuration making up the base schema.
|
Table |
View |
Description |
|---|---|---|
| tb_event | vw_event | (Highly Recommended) Dataset built by Event Enrich process at the level of individual events (aka Hit or Record) that is a foundation for building all other Composer AI models and analytical datasets |
| tb_visitor_daily | vw_visitor_daily | (Optional) Dataset built by Visitor Enrich process that groups together the individual tb_event records by visitor identifier codes and event_partition, and uses aggregation functions such as min(), max(), first(), sum(), and last() for each field |
| tb_session | vw_session | (Optional) Dataset built by Session Enrich that groups together the individual tb_event records by session_id and uses aggregation functions such as min(), max(), first(), sum(), and last() for each field |
| tb_product | vw_product | (Optional) Dataset built by Product Enrich that is configured for analysis of orders by individual product codes, such as unpacking the Adobe product list as individual rows for each product code stored within the array |
| id_graph | (Not Applicable) | (Optional) Dataset built by the Register Identity process that searches through the input sources for all combinations of local_id and universal_id pairs. This table is required for Unified Event Store app, but otherwise optional. |
Base dataset supporting tables
Upon initialization, the above Syntasa behavioral schema tables are built. There are various supporting tables that are also built to assist in building the Syntasa behavioral schema. Below describes these other tables that are found within an Adobe data source.
|
Table |
Description |
|---|---|
| browser, operating_system, operating_system_type, search_engine, etc | Adobe provided lookup files that are included within the raw data feed files. These tables are joined in the SYNTASA behavioral schema. |
| raw | External table created on the as-is raw data files stored in the customer environment. This table can be queried for data debugging purposes, but is not recommended for normal analytical use as it is massive resulting in slow query times, and the data is not flattened nor standardized. |
Analytical datasets and lookups
Syntasa does provide the ability to create custom datasets and to process lookup files. These do not have a standard naming scheme and could be written to any defined Event Store, therefore, it is critical to be aware of this and work with Syntasa services professionals to define an organizational best practice for the creation of these datasets. Best practice may include a set of rules including naming convention, approval process, and recommended structure.