Internal database schema
The following tables are used internally by DeepDive. There can not be tables with the same name in the database:
List of relations
Schema | Name | Type
--------+----------------------------------------------+----------
public | dd_graph_variables_holdout | table
public | dd_graph_variables_observation | table
public | dd_graph_weights | table
public | dd_inference_result_variables | table
public | dd_inference_result_weights | table
public | dd_inference_result_weights_mapping | view
public | dd_factors_[RULE_NAME] | table
public | dd_weights_[RULE_NAME] | table
public | [TABLE]_[VARIABLE]_inference | view
public | [TABLE]_[VARIABLE]_calibration | view
public | dd_categories_[TABLE] | view
where [RULE_NAME] is the name of an inference rule, [TABLE] is the
name of a table that contains variables, and [VARIABLE] is the name
of a variable in the corresponding table.
Description of each schema:
dd_graph_variables_holdout: a table that contains all variable ids that are used for holdout. Can be used for custom holdout by a holdout query.dd_graph_variables_observation: a table that contains all variable ids that are evidence that will not be fitted during learning. An usage example of this table can be found here.dd_graph_weights: a table that contains all the materialized weights.dd_inference_result_variables: a table that contains the inference results (expectation) for all query variables.dd_inference_result_weights: a table that shows factor weight ids and learned weight values.dd_inference_result_weights_mapping: a view that maps all distinct factor weights to their description and their learned values. It is a commonly used view that shows the learned weight value of a factor as well as the number of occurences of a factor.dd_factors_[RULE_NAME]: a table that is defined by the input query of an inference rule. You can use it as a feature table in BrainDump.dd_weight_[RULE_NAME]: a table that stores initial weights for factors, used internally.[TABLE]_[VARIABLE]_inference: a view that maps variables with their inference results. It is commonly used for error analysis.[TABLE]_[VARIABLE]_calibration: a view that has calibration statistics of a variable. Used in generating calibration plots.dd_categories_[TABLE]: a view that records the cardinality series of given variable. For example if the domain of a variable is {1,2,3} the cardinality is 3.