Complex Friend Recommendation

App Development
1

I am pretty much convinced about Dgraph and started my development of a social network using it. Currently, I am a bit confused about some query structure. Some help will be highly appreciated.

My User Schema is

type User @secret(field: "password") {
    uuid: String! @id @search(by:[fulltext])
    email: String! @search(by:[fulltext])
    location: Point @search
    presentAddress: Point @search
    permanentAddress: Point @search
    friends: [User] @hasInverse(field: friends) 
    contacts: [Contact]
}

type Contact {
  id: ID!
  name: String! @search(by:[fulltext])
  number: String! @search(by:[fulltext])
  email: String! @search(by:[fulltext])
  phone: String! @search(by:[fulltext])
  user: User @hasInverse(field: contacts)
}

I am designing a friend Recommendation query based on the current location, present address, permanent address, and contacts that I have in my database along with the friends of friends.

so I end up with the following DQL

{
  var(func: eq(User.email, "email_address")) {
    myscore as  math(5) 
    User.friends {
      isMyFriend as math(1)
      User.friends{                
        fscore as math(myscore)
      }
    }
  }
  var(func: near(User.location, [90.412521,23.810331],100)) {
    locationScore as  math(5) 
  }
  
   var(func: near(User.presentAddress, [90.412521,23.810331],100)) {
    presentLS as  math(5) 
  }
  
  var(func: near(User.permanentAddress, [90.412521,23.810331],100)) {
    permanentLS as  math(5) 

  }
  ## I NEED A CONTACT MATCHING PART REMAINING
  var(){
    fofscore as math(fscore)
    currentLocationScore as math(locationScore)
    presentLocationScore as math(presentLS)
    permanentLocationScore as math(permanentLS)

    SCORE as math(fofscore+currentLocationScore+presentLocationScore+permanentLocationScore)
  }


  
  suggestedFriends(func: uid(SCORE), orderdesc:val(SCORE), first:10, offset: 0)
    @filter(
      and(
      not eq(val(isMyFriend),1)
      not eq(User.email,"email_address")
    )
  ){
    
    score : val(SCORE)
    User.uuid
    User.email
    # expand(_all_){
    #   expand(_all_)
    # }


  }
}

Now I only need contact matching part
Any suggestion will be highly appreciated :blush:

@pawan @ashwin95r @MichelDiz

2

@pawan any suggestion