Extension:ORES/Schema
ores_model
editThis table keeps data about models and their versions.
mysql:wikiadmin@db1089 [enwiki]> DESC ores_model;
+------------------+----------------------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+------------------+----------------------+------+-----+---------+----------------+
| oresm_id | smallint(5) unsigned | NO | PRI | NULL | auto_increment |
| oresm_name | varbinary(32) | NO | MUL | NULL | |
| oresm_version | varbinary(32) | NO | | NULL | |
| oresm_is_current | tinyint(1) | NO | | NULL | |
+------------------+----------------------+------+-----+---------+----------------+
4 rows in set (0.00 sec)
oresm_id
editPrimary key, oresc_model is a foreign key to this column.
oresm_name
editName of the model, like "damaging", etc.
oresm_version
editVersion of the model, like '0.1.1' or etc.
oresm_is_current
editWhether the model at this version is current version in the service or not.
ores_classification
editThis table keeps scores of each revision in model and class
mysql:wikiadmin@db1089 [enwiki]> DESC ores_classification;
+--------------------+---------------------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+--------------------+---------------------+------+-----+---------+----------------+
| oresc_id | bigint(20) unsigned | NO | PRI | NULL | auto_increment |
| oresc_rev | int(10) unsigned | NO | MUL | NULL | |
| oresc_model | smallint(6) | NO | | NULL | |
| oresc_class | tinyint(4) | NO | | NULL | |
| oresc_probability | decimal(3,3) | NO | | NULL | |
| oresc_is_predicted | tinyint(1) | NO | | NULL | |
+--------------------+---------------------+------+-----+---------+----------------+
6 rows in set (0.00 sec)
oresc_id
editPrimary key
oresc_rev
editForeign key to revision_id in revision table.
oresc_model
editForeign key to oresm_id in ores_model table.
oresc_class
editclass that ores gives the score. 1 for true and 0 for false in binary models (such as damaging and goodfaith). More numbers for more classes in multi-class models (For example in wp10, 0 is for "Stub", 1 for "B" and so on)
oresc_probability
editProbability that ORES services sends out, rounded to 3 decimal digits for performance.
oresc_is_predicted
editWhether ORES service thinks this is the most fitting class or not. If a class is binary, when probability is more than 50% this is 1 but for multi-class models, it's more complicated.