Neptune query examples to execute during this demo

[deejvince]

Hi,
I'm looking for instructions on how to query Neptune using gremlin and display some things like "if you boughts this, you'd also want to look at that"
Can you share some examples that can be demoed using this application?

[triggan]

Hi Lior,

As of right now, the dataset that we used to boot strap the Neptune instance only has customers that have purchased a single item. In order to do the type of traversal and recommendation pattern that you suggest, the initial dataset would need to be expanded such that users have purchased more than one item. Then you could make recommendations based off of what other books people were purchasing from the view point of one particular book.

The current recommendation pattern shown in the app makes recommendations based off of a fabricated social network. We established friends between users and made recommendations based off of those 'friendOf' relationships.

Once the graph dataset were expanded and further transactions were added, you could use the following traversal to make recommendations based off of a single book:

g.V("2rb37qw5-d93b-11e8-9f8b-f2801f1b9fd1").as('a')        #Start with "Behind the Symbol"
    .in('purchased')        #Find all people that purchased the book
    .out('purchased')        #Find all books that those people purchased
    .where(neq('a'))       #Make sure you're not including the original book
    .valueMap(true).dedup()       #Output attributes of verticies returned and dedupe

This example shows a traversal for book recommendations for the book "Behind the Symbol" in the sample dataset. Like I mentioned, this will unfortunately return 0 results due to the current content of the initial dataset.

We're open to PRs for extending the existing dataset to make this more interesting.