Search
Clear search
Close search
Google apps
Main menu

How can I tell if ASR user clicks (self learning scorer) is working correctly.

Summary: ASR has been enabled but unsure if it is working correctly.


Cause: Even if ASR is enabled the biasing can not be observed easily, it is not possible to clearly identify it's effect or relevancy.


Troubleshooting Steps: First please ensure that ASR is infact enabled on the relevant front end and that you have enabled a biasing policy under Google Search Appliance > Serving > Frontends > Filters .
To determine if ASR is working, first perform a search and click on some results, preferably using several different browsers on different IPs, then export an advanced search report as described below here, ensuring you do it for the correct collection. The advanced search report should list relevant data pertaining to user clicks, this shows that the GSA is detecting user clicks.
To see the effect on a search result the GSA needs to be in a live environment where users are searching and clicking on results, in such an environment perform a search then remove the proxystlesheet parameter from a search URL and add the entsp parameter for your biasing policy as described here in the xml reference. eg. for default_policy entsp=a
The clickboost bias on results is part of the SCOREBIAS tag as described in the xml reference A clickboost score bias should be visible in the results if clickboost has been applied to your URL. eg. name="ClickBoost" value="1"

Please note that click boost data can not be easily "faked", it works best in a real environment where users from different locations are performing realistic searches.


Additional Information:  Using the Automatic Self-Learning Scorer

Was this article helpful?
How can we improve it?