This section has some thoughts on future work, improvements and next steps. Please feel free to PR your ideas and suggestions.
In previous versions of the lab we had users sign up for a new Azure account they owned. That was kinda cool in that attendees got to see everything start from scratch. However, the signup required a credit card number and a phone number for identity verification. It was also a fair bit of work. Now we're using OneBlink.AI to provision. We'd be curious to hear how your experience was with this approach.
Other approaches provide a way to access Azure. However, the result is different from what a user experiences in a real environment. The goal of this lab was to give you hands on end to end experience. Given that, this seems to be the best way available.
We deployed an IaaS listing for Neo4j. That uses an ARM template internally which has a variety of pre-requisites. Neo4j is currently working on a click to deploy version of our managed service for Azure called Aura. It's available today but does not yet support Graph Data Science (GDS). Once it does, we'll transition this material.
You can view options for deploying in the Azure Marketplace.
We connected over HTTP. Ideally we'd use HTTPS. The setup is complex however and the UX ideal because of self signed certificates. Using Aura avoids this issue entirely.
We used LOAD CSV to pull data in. That is one of many ways. Neo4j Data Loader is another.
You can use some of these connectors from Neo4j to get data in:
Neo4j Graph Data Science also supports PyArrow which can make data transfers in the GDS much faster.
This section of the lab could be greatly expanded.
We only parse 3 files here. That's down to very limited quotas. With more, we could make a bunch of parallel calls. It'd be neat to get to that point and parse all the data that way rather than with LOAD CSV.
The chatbot is somewhat brittle. More work could be done to improve it. You can almost certainly think up some questions that it should answer but can't. That's part of what is so exciting about this space -- everything is developing quickly.
The account we're using has limited quotas. That forced us to throttle.
We hope you enjoyed these labs. If you have any questions, feel free to reach out directly to any of us. We'd love the opportunity to explore and support your use cases with your data.