search

actionml / universal-recommender

Highly configurable recommender based on PredictionIO and Mahout's Correlated Cross-Occurrence algorithm
http://actionml.com/universal-recommender
Apache License 2.0
669 stars 172 forks source link

Location based recommendations #66

Open NaikSoftware opened 5 years ago

NaikSoftware commented 5 years ago

Can I take into account user/item locations in universal recommender? If it is true how to do it? Thanks.

mick912 commented 4 years ago

I think it will work input:

{
   "event" : "$set",
   "entityType" : "item",
   "entityId" : "iPad",
   "properties" : {
      "category": ["electronics", "mobile"],
      "country": ["UK"]
      "expireDate": "2016-10-05T21:02:49.228Z"
   },
   "eventTime" : "2015-10-05T21:02:49.228Z"
}

query: 

{
  "rules": [
    {
      "name": "country"
      "values": ["UK"],
      "bias": -1
    }  ]
}
zebraf1 commented 2 years ago

Yes, you can add a property to item which has array of values. When quering you can filter only items which have this property value using bias -1 or boost it's value by using bias > 1. Using a value between 0 and 1 deboosts it and using 0 excludes it.

I this this issue should be closed.