 Okay, good with the timer. Yep. Yep. Take it away. Okay. My name is Jones as mom See your friends in can we have a database called a LEGO graph and IDF triple store? And we get a lot of customers that say well, you're called a graph database But you're not really a graph database because you don't do property graphs So it's it's something that I want to put an end to so I made this very tiny little presentation So we're doing a project for a big bank that saves the bank about 30 million dollars a year in fraud detection Where we look at all kinds of interesting patterns that might indicate some kind of collusion or other fraud One query could be okay find a circle of payments save it in some of within 10 miles of son or say Within the last day with the payments are more than a thousand dollars. Yeah, so you mixed you spatial temporal and all those other things Now normally if you would do this you would do it in a graph database You would have nodes links between nodes and what makes a property graph is that then the links between nodes also have properties Yeah, so here you see that this person pasted oops this person paste this person but This pace relationship have some properties. This will have other ones etc. Etc. Now in most great graph databases This won't really skill because it's an aggregate query Yeah, the graph databases work fantastic if you start with our node and you do a fan out If you want to do a big aggregate query Then you're better off is something that like works like a relational database or at least the principles of relational databases And then most graph databases don't do much of geospatial and temporal So what we did is we in our graph database we implemented The property graphs and hypergraphs and then those property graphs we can you do with geospatial and temporal indexing And then our query engine can deal both with the graph search and with the property graph and for anyone that knows about RDF Who knows about RDF is in this audience? That's cool So this is the way where you do it where you use the fourth element of the triple to create a handle where you then hang everything Everything off and actually in our approach you can even have property property property graphs because you can keep going all right, so Then once you have this I can do queries in the w3c language for RDF graph databases called sparkle and here's a query where we try to find a ring of payments where the first part of the query Will try to find the circle and When we find the circle we also find the handles for the the payments And then we can look at the property graph for each of these links And then we can say well it has to be today and it has to be ten miles of Son of say Yes, I can demo this and I'll just show you how it looks like. I've still two minutes left. Oh Gosh, where's my link? Here it is So we have this database put together and then we created this query language where we actually can do graph search Without writing any code. So basically I can take I can make variables. I can make links between variables and Then so here I've tried to find a ring between a and B and C and D and E and then back to a And then just for fun. I want to have the property graph for this particular link So I get this here and then I just can say turn this into a sparkler prologue query in this case. I choose sparkle I Do the query and you see This visual graph is turned into a sparkle query here to get the results And then I can visually inspect the graph to see what's actually going on. Does it make sense all right? So going back to my presentation So this morning I actually talked about How we used Hadoop to extract data from this bank to put it into a LEGO graph and do analysis We're now working with Intel to to make it easier to get these graphs out of Hadoop and tomorrow I will talk more about my Geospatial indexing and especially when you have moving objects in your database How can you find these objects as soon as possible or as fast as you want? Okay? Thank you very much