The **Recommendation** feature provides a simulated environment to help you verify and debug the actual effectiveness of your recommendation strategies before going live. This ensures that the results align with your business expectations. It is applicable to various scenarios, including Homepage and Detail Page recommendations. This guide will walk you through how to simulate real\-world business requests by configuring test users, items, and geographic locations to preview recommendation results. ### Workflow 1. Access the recommendation experience center: navigate to "Experience Center" \> "Recommendation" in the console, and select the target recommendation configuration (such as homepage recommendation or product listing page). 2. Simulated input configuration: the platform supports simulating users, user geographic locations, and parent Item. --- ## Configure test users You can configure test users through **specified user mode** or **random user mode**. **Random user mode** selects a user ID randomly by the system for quick recall effect testing; **specified user mode** requires entering a test user ID, whose profile will display automatically and be used for personalized recommendations. ```mixin-react return (

3. After clicking "OK", the user profile displays automatically. Recommendation results update automatically based on the new user profile. **Hover over "User interest tags"** **to view tag strength.** `}>

2. The platform randomly selects a valid user ID from the behavior data set. Recommendation results update automatically based on the new user profile. **Hover over "User interest tags"** **to view tag strength.** `}>); ``` ### Configure test items In the **product detail page** recommendation configuration, personalized recommendations also use test items as parent/main item to obtain related items. The attributes of test items will be used for personalized recommendations. ```mixin-react return (

3. After clicking "OK," the item details will display automatically. Recommended results will also update automatically based on the new test item. `}>

2. The platform will randomly select an item from the item dataset. Recommended results will also update automatically based on the new test item. `}>); ``` ### Set user geographic location After configuring **boosting and demotion rules ** related to geographic location, you can verify the rule effects in the recommendation experience interface through the following steps to intuitively experience recommendation adjustments based on user location (LBS) and item distance. :::tip \*Prerequisites At least one "Geolocation Rule" has been added; see details: [ ](/docs/85296/1963520). ::: 1. **Enter the recommendation experience interface** Navigate to "Experience Center" \> "Recommendation" in the console and select the target recommendation configuration/scene (such as homepage recommendation or product list page). 2. **Simulate user geographic location and item distance** In the right "Test User" area, use the location features to simulate scenarios where a user's geographical position triggers boosting or demotion rules:

|**Capabilities** |**Capability Description** | |---|---| |Get My Location |Obtain your current location to simulate user testing. This requires you to grant location permissions to the browser.|\ | |:::tip \*Automatic Acquisition Failed|\ | |If clicking "Get My Location" does not automatically retrieve latitude and longitude, please check:|\ | ||\ | |1. Whether the computer has location services enabled|\ | |2. Whether the computer or browser restricts browser location access permissions|\ | ||\ | |If the browser's location permission was previously denied, you must restart the computer to reset the permission restrictions.|\ | ||\ | |:::| |Specify User Location |Manually enter latitude and longitude coordinates (e.g., `Longitude: 39.9°`, `Latitude: 116.4°`) | 3. **View changes in recommendation results** Click "Recommend again". The system will apply configured boosting/demoting rules based on the simulated user location and the actual distance of items (requires item dataset with latitude and longitude configured). For example: * **Boost effect for nearby items**: if the rule sets "boost items within 5 km of the user by 50%," items within this range will rank higher in the recommendation list. * **Demote effect for distant items**: if the rule sets "demote items beyond 20 km from the user by 80%," items beyond this distance will rank lower in the recommendation list (or receive less exposure).

Viking AI Search