Closed narges1212 closed 3 years ago
If a session-identifier is provided, the same systems should be queried in order to return consistent results to site users within a specific session. Currently, we do not cache the rankings/recommendations. In fact, this will result in a new request of the experimental system which consequently leads to a new rid
.
The idea is to make a new database entry for each ranking and recommendation even if they contain the same items. Each time we display results to users, their interaction and thus the logged feedback can be different. Long story short: it is intended to return a new rid
.
Btw: the endpoint datasets?itemid=<string>&sid=<int>
does not offer the option to pass a rid
. Maybe this has caused the confusion!?!
Hi Narges, please have a look at this commit. I've added an endpoint to retrieve older rankings and recommendations with the help of rid
. The corresponding endpoints are:
/recommendation/datasets/<int:rid>
/recommendation/publications/<int:rid>
/ranking/<int:rid>
Please tell me, if the formatting of the output is okay!
Thx!
When testing, please have a look at the unit test I've added in this commit
1- /recommendation/datasets/
all return the same result, which is correct, but I am not sure about the difference btw 1 .. 3 the returned json contains always "feedback_id"=null (what is the usage of feedback_id?)
This means, there has not been feedback data provided yet.
ok, for now all returns same result. tested! done
To prepare the feedback I need to send again a get request to stella-app to get the ranking with specific ranking_id, however I got always a ranking with new rankig_id. the session id and item_id are returned correctly.