Design for Paired Comparison

GSQL Graph Schema
#1

We need to graph a “Paired Comparison” and could use some recommendations on best practices for this.

Scenario:

  • A “User” must compare 2 items against each other and choose 1.
  • When making the choice the user must enter a rating: Was this a easy, medium or difficult choice to make.
  • The user will enter a short explanation describing why he/she made the choice.

Current design:

The data for the 3 bullets above will be in a vertex called “Evaluation Event”

The screenshot below shows how we have mapped this out.
image
If there is a total of 6 items to compare, there will be 10 comparison events.

And the graph would look something like this.
image
There is a vertex between Item and EvaluationEvent called “Item2EvaluationEvent” where the source vertex is “Item” and the target vertex is “EvaluationEvent”.
In addition to this, we’re writing the IDs of the Items to the attributes “Chosen_Item_Id” and “NotChosen_Item_Id”.

Is this good graph design and is there a better way to do this?

Thanks.

#2

image
image622×561 57.4 KB

@gpadvorac Looks good. You can store the “choice” the user made on the “CHOICE_HAS_ITEM” in an attribute called “selected” and have that as a Boolean (true or false)

USER_HAS_CHOICE

  • timestamp

Event

  • rating
  • reason

User

  • id
  • name
#3

We could have separate events to encode the choice/not-choice e.g. a “chosen in” type edge (from item to event), and a “not chosen in” type edge (from item to event), thus encoding the choice explicitly in the edge, rather than an attribute.

That would make querying more explicit and be a little more idiomatic in that transitive verbs are modelled as edges.