People use Neo4j to manage enterprise architectures all the time. If you haven’t seen this presentation from Thomas Lawrence from Amadeus, then you owe it to yourself to watch it. But what about lower level networks? Can we use Neo4j to model routing in a physical network? Of course we can, and today I’ll show you how.
Did you know that Chicago is the most important railroad center in North America? Chicago has more lines of track radiating in more directions than from any other city. The windy city has long been the most important interchange point for freight traffic between the nation’s major railroads and it is the hub of Amtrak, the intercity rail passenger system. You may not realize it, but railroad tracks and graph theory have a history together. Back in the mid 1950s the US Military had an interest in finding out how much capacity the Soviet railway network had to move cargo from the Western Soviet Union to Eastern Europe. This lead to the Maximum Flow problem and the Ford–Fulkerson algorithm to solve it.
When I joined Neo4j, I didn’t know how to write Java. I was a SQL developer who knew some Ruby and that’s about it. Luckily I had Michael Hunger, Stefan Armbruster, David Montag and others to help me out. I realize however that you may not be so lucky. So today I’m going to share with you a set of slides to help you start you on your journey of using the full power of Neo4j.
Continue reading
So far I’ve only showed you how to traverse a decision tree in Neo4j. The assumption being that you would either create the rules yourself from expert knowledge or via an external algorithm. Today we’re going to add an algorithm to build a decision tree (well a decision stream) right into Neo4j. We will simply pass in the training data and let it build the tree for us. If you are reading this part without reading parts one, two, and three, you should because this builds on what we learned along the way.
At Graph Connect this year I did a short lightning talk on building Decision Trees using Neo4j. The slides are up and down below, the video is up. After the talk, someone asked, “What if we don’t know all the facts ahead of time?”. They wanted to be able to step through the tree and ask for the facts as needed at each step. So today we’re going to see how to do that.
Continue reading
In Mr. Rogers’ Neighborhood, the question “Won’t you be my neighbor?” is an invitation for somebody to be close to you. In graphs, it’s an invitation to traverse. The closest neighbors of a node are those reachable by a single relationship hop, but we can also consider nodes two, three or more hops away our neighbors as well. How can we find them in Neo4j? Using the “star”:
Continue reading
The multiple-origin-multiple-destination (MOMD) problem is an NP-Hard problem sometimes seen in logistics planning where paths can stretch out really far. A far simpler problem presents itself when we limit the size of the paths. Now you may be wondering, why would we do that? Well… outside logistics we have plenty of graphs where relevance drops as we get further and further away. Think about an Article on Wikipedia. It has links to many other articles that are relevant, and those have links to other articles that are relevant to them but less relevant to our starting Article, and those have links to other articles that may be relevant to them, but have very little to do with our starting Article. I think if we keep going we end up in Philosophy or something like that.
Continue reading
It’s time to add “visions of love” to our dating site. So far our posts have been just text status updates and while it is possible to fall in love with someone’s words, it’s harder if they look like the troll that lives under the bridge. So what’s the plan here? Well… like most databases out there, it’s not a good idea to store images in Neo4j. What we are going to store instead is a link to where the image resides… but we also don’t want to deal with having images all over our file system and then having to worry about storage space and replicating them, geographically distributing them for faster access, etc. Hosting images is a problem solved by the use of Content Delivery Networks. So let’s leverage one and build our feature.
Continue reading
Up to this point, our users can send and receive messages, but we don’t have a way to show them all of their conversations, only one conversation at a time and they have to guess who messaged them before they can see those. Not very useful, what we need is a directory of all the conversations our user is part of. Let’s go ahead and add this feature to tie things together.
Continue reading
A few days ago I taught the Neo4j Modeling class in Dallas… well my own version of the class since I teach some of the older material plus some of my blog posts. If you ever get the chance, take the class, it will open your eyes at what is possible when “third normal form” gets thrown out the window and a whole new world of possibilities is before you. The next modeling class I’m teaching will be at Graph Connect in NYC on September 20-21, 2018. I recommend you join me, unless of course you are looking at this after it occurred, then catch the next one.
Continue reading
