For every recommendation scenario a set of filters can be defined that is only valid for the specified scenario. They are applied to all recommendations from every model linked to the scenario. Below you can find the user interface to manage filters and a detailed description.
|Filter||Requirements and restrictions|
|Remove already purchased products|
The recommendation engine stores the buy and consume events for every user (anonymized) for one year. If this filter is activated the user won't get this products recommended again. More about events in the chapter 8. Event Types.
This filter is available for both basic and advanced solution.
|Do not recommend the same product multiple times||If this filter is activated the product will be removed from all recommendation lists of the current user session, if it was already recommended the defined amount of times.|
|Do not recommend the product currently viewed||Remove the context items from the recommendation list. Usually it makes sense to activate this filter except in the "Ultimately bought" case.|
Filters which are available only in the advanced solution
|Product price should be higher than the specified value|
This filter could be used to remove very cheap but popular items from the recommendation list. For example an optometrist wants to show the most popular design glasses on the home page to avoid displaying insurance subsidized cheap models and glasses cleaning cloths.
This filter uses prices exported to the recommendation engine (see 11a. Item import). Prices attached to the BUY event is not used here.
|Product price should be equal or higher than the price of the context product||As the filter above it compares prices exported to the recommendation engine (see 11a. Item import).|
|No top-selling products|
Removes products from the recommendation list which are recommended by the top selling model (even, if the model itself is not linked to this scenario). It is another way to remove very popular products from the recommendation list if you are not interested in selling them again.
Of course it makes no sense to apply this filter to a top selling scenario. It will filter everything out.
|Special always enabled filters (advanced edition only)|
|Editor blacklisted products.||Advanced edition gets several editor based models (see chapter 5. Recommendation Models). One of them is an editor blacklist. If this filter is activated, the products from the black list will be removed from every list of recommendation. One can use this feature to remove some standard products which users will buy in any case without recommendation (bread, butter, milk) or some non existing test-product which is only used by a robot to check, if the shop software is working properly.|
|Customer blacklisted products||Additionally to the standard set the website can send blacklist events for selected products and users (see chapter 8. Event Types). It can be implemented as a button "I do not like this product". If a user added a product to his black list, it will not be recommended anymore.|
|Product validity interval||Additional product information can be exported to the recommendation engine (see 11a. Item import). If the product information is available and contains a validity time interval (for example the product is available only until 24.December 2014), it will be used to remove non valid products from the recommendation list.|