Blender is for combining different item lists into one. There are several options for combination by defining the number of items taken from each input, in the order of input and the combination strategy.
Blender can be used e.g. to set up a Fallback Logic. If 10 items are requested, than set the “Limit item count” to 10. Set the primary logic as the first input and limit it to 10 as well and set the Fallback Logic as a second input and set the limit to 10 too. This way if the primary logic can provide at least 10 items, all the items will be equipped by the primary logic. Otherwise, the empty places will be filled up by the Fallback Logic. Fallback Logic do not necessarily have to be a different algorithm, e.g. it can be the same list of items just without filtering as on the figure below.
On other use-case is to mix algorithms, in this case the limit for each block should be less than the total limit. E.g. both block limits can be set to one, connect the two blocks with each other by drag-n-drop and turn on repetition.
Combinator Node is for combining the p-values of two predictors. E.g. the p-values of a personalized logic and a popularity logic can be added. For summing up the p-values any kind of mathematical formulas can be used. In the example below the first input will have twice as much weight as the second.
Moreover, item properties and special functions like IF can be used in the “Expression”.
Updated about 3 years ago