Reltio Connect

One of the things I have used in the past, is that we have an open-source tool for potential match report, and I can put a link to that in the chat here in a second. But it's basically a little Java app that'll spit out two records side by side, and you can compare externally. You're going to need to make the matches and merges in Reltio to bring the record together. Someone is going to have to come back and match and merge. I think there's some cool things, another way you might build a handle of that, if you want to have a downstream system making these decisions, is when two things become potential matches, you can have it create a message in your SQS queue. And that downstream system could take those messages, create a queue for a data steward to go through, and you could create an API that handles the merge. There's an API that you could call to merge two records

There are a few different ways you can go about that. Let me put this data extract tool in the chat, so that folks can find it later to look at their potential matches. Speaking of tuning in and looking at your matches after the fact, these first four slides were really driven about how do you build your rules, what should your roles look like? The next couple of slides I want to talk about, now that you have built your rules, how do you tune them? And the first tool that's built into Reltio, it's nice, it generates this HTML report, static match rule analyzer. And what this thing is doing is it's checking the syntax and your comparator and tokenization classes to make sure basically you configured the rule right. This is data agnostic, you could have no data, you could have a ton of data, it's not considering your data, it's purely looking at your rules and giving you feedback in this comment section on whether your rules are designed well, syntactically correct, redundant, that sort of thing.

And there's a almost a stoplight coloring system to tell you what's going on with your rule. If it's green, your rule is well configured and pretty much ready to go from a syntax perspective. If it's yellow, it might be a duplicate rule, they have a lot of overlap, and you might want to consider combining them or making them more specific. And blue means, hey, your match tokenizer, your token choice, and your comparator choice, aren't matching up and you need to review that. And I know that's a little vague, so if you look at the comments section, it'll give you more details about which choice you maybe should have made instead. So let me show you how you generate this report. From the console, you'll go back to the data modeler where you're configuring your match rules. And from here, you can hit analyze, and it's going to generate a nice little HTML report and ask you to save it.

You can open it up and see what we'll see here in my demo tenant, is we have a few rules that have very similar strategies on matching for first and last name. And it's calling those out here, that we're using the same tokens to do fuzzy in both rules, maybe we want to double check that's how we want to handle it. So that's how you generate that report, because this is an out of the box rule set, there's not a lot of comments because things are pretty much well configured, but after generating your own rule in the match role builder here, you're going to want to always do that analyze tool to get an idea for if you configured it properly, before you start using it to match and merge.

Find a community webinar on Match and Merge!