You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/personalization/recommendation_models.md
+30Lines changed: 30 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -214,6 +214,36 @@ The logic used for resolving a submodel is as follows:
214
214
Recommendation are fetched from all the submodels and merged based on the weight (relevance).
215
215
If one of the submodels delivers recommendations with better relevance, the results of other models can disappear from the list.
216
216
217
+
### Dynamic attributes
218
+
219
+
Dynamic attribute submodels eliminate the need for manual grouping and simplify configuration.
220
+
They allow for simpler, faster, and less demanding recommendation models building using different attributes, because all you need to do is make one request and rebuild the model.
221
+
222
+
They work best in straightforward cases when you filter by the value of the attribute.
223
+
224
+
Dynamic attribute submodels:
225
+
226
+
- operate only on [nominal attributes](recommendation_models#nominal-attributes) (numeric attributes are not supported)
227
+
- can be used for [popularity](recommendation_models#popularity-models) and [collaborative](recommendation_models#collaborative-models) types of models (as they support submodels)
228
+
- have limitation of max. 50 attribute values (if more, you need to follow the procedure of manual configuration by [[= product_name_base =]] Team)
229
+
- operate on scenarios with the selected `Submodels` data type option
230
+
- require sending a request and building a model
231
+
- are calculated for all new attribute values after import
232
+
- are always up-to-date with the imported items
233
+
- still add new values when attributes are only partially grouped manually
234
+
- aren't added if all attributes are manually grouped (full manual intervention)
235
+
- cannot be calculated if there is any submodel manually configured for provided attribute
236
+
- don't operate on the values which are no longer present
237
+
238
+
!!! note "Enable dynamic attribute"
239
+
240
+
Dynamic attribute must be enabled by [[= product_name_base =]] Team.
241
+
To start using this functionality, contact customer support ([email protected]).
242
+
243
+
!!! caution "Unused attributes"
244
+
245
+
If an attribute is not used for at least 5 days, all related submodels are removed.
246
+
217
247
## Time-slot based models
218
248
219
249
Time-slot based models consider only a particular range of time rather than the full day when calculating recommendations.
0 commit comments