Emtech Group Podcast
Episode #3: Unveiling the Power of Knowledge Graphs
Summary
In this episode of the Emtech Group Podcast, our guests delve into the transformative potential of knowledge graphs when applied to quality assurance across the insurance industry value chain. They draw on diverse experiences to highlight the parallels between traditional methods, such as flow charts in IVR systems, to leveraging knowledge graph for quality assurance automated testing. Their discussion emphasizes the move towards the use of knowledge graphs for comprehensive coverage and flexibility in testing workflows. They highlight the benefits of knowledge graphs in enabling collaboration across different teams and domains, without requiring specialized programming knowledge.
The conversation underscores how knowledge graphs facilitate the ability to map complex flows and traverse them rapidly, driving automation into test case and data generation, while the visualization makes it easier for non-developers to understand. Despite not being explicitly labeled as such in the past, Jeff and Spencer demonstrate the value of the knowledge graph approach in streamlining processes and fostering collaboration.
To find more episodes of Emtech Group Podcasts, visit our resource center. To read more about Quality Assurance, QMT, and technology topics, visit our blog.
To find the transcription of this podcast, scroll to the bottom of the page.
The views and opinions expressed in the podcast are those of the guests and do not necessarily reflect the official policy or position of Emtech Group Inc.
Featuring
Jeff Kramer
Jeff is a is a Senior Software Development Manager at Emtech Group. With over 30 years in the software development and services industry, Jeff has accumulated a wealth of experience across diverse fields such as high-performance scientific computing, real-time interactive audio and video systems, and e-commerce. Throughout his career, he has held pivotal roles in the development and operations of mission-critical systems, where the stakes of downtime and errors are high.
Spencer Reuben
Spencer is a Program Manager at Emtech Group. He is a proven industry leader with deep specialization in software testing, software quality assurance, and IT systems integration. He has many years of proven IT leadership, specializing in designing automated software test solutions for insurance carriers, automotive, and robotic industries, leveraging non-traditional approaches such as model-based testing, AI/ML.
Transcript
00:00:17 Jeff Kramer
Hey, Spencer
00:00:19 Jeff Kramer
I was sitting here thinking on this cloudy, wonderful winter Toronto afternoon about the fact that I actually had worked in an industry.
00:00:28 Jeff Kramer
Me that although the terminology is different, was using knowledge graphs years ago.
00:00:35 Spencer Reuben
Oh, really? That sounds cool.
00:00:36 Jeff Kramer
Yeah, it was the telephone. It’s well, the term is IVR interactive voice response, but it’s those systems that you interact with whenever you try and call a bank or an insurance company or you want to talk to a live person.
00:00:51 Jeff Kramer
And you must walk through the various menus and recordings in order to get stuff done. Ohh cool. Yeah, and it’s surprising that they’re actually that. There is this commonality there. And I it was not expected, but the company that I worked for used to actually model their telephone script.
00:01:11 Jeff Kramer
Workflows with flow charts. That was the terms that we used.
00:01:16 Spencer Reuben
OK.
00:01:16 Jeff Kramer
And you don’t necessarily see the similarity at the beginning, but it clicked when I started taking a look at what our application does, what QMT does, and there are a lot of similarities. OK, cool. It’s more modern. You’re interacting and modeling interactions with a different technology, but in the end.
00:01:37 Jeff Kramer
Part of what you’re modeling with QMT is basically how a user is going to interact with the various systems that the insurance company uses in order to.
00:01:48 Jeff Kramer
Set up a policy or do one of the other parts of the workflow and a lot of that is basically user does this. Now what’s the response supposed to be presented in the graph that makes it nice and simple and easy for somebody to understand and the telephone the IVR industry had a similar sort of thing.
00:02:08 Jeff Kramer
And they, their solution? This company was a was an analogous solution to this, and it’s really.
00:02:14 Jeff Kramer
Praise me that there were the similarities there.
00:02:17 Spencer Reuben
I see. You know what that’s really interesting because you know, I come back. I come from a long-time testing background, right? So, software testing, you know, and a lot of things were started. Very manual, you know, 20 some odd years ago. You know, we did automation.
00:02:34 Spencer Reuben
But automation really was. I found it was very tricky, right? You did have to have a developer out on. You did have to get down on the weeds. You did have to get the code base. And if you didn’t have that skill.
00:02:47 Spencer Reuben
That you would get a set of automation tests working. You would get help from your development team and then they would never be maintained. So it was, uh. It was a very tricky thing to keep the automation maintained.
00:03:00 Jeff Kramer
Yeah, that’s actually the maintenance thing is, is actually key in the in the use of developers. I mean my background is primarily doing.
00:03:09 Jeff Kramer
Development and when we worked there, they still had developers doing a lot of the development of these IVR systems, but they had non developers doing a lot of the design and doing the development of the call flow because the tools that they had using their flow chart basically allowed them to.
00:03:29 Jeff Kramer
Drag and drop.
00:03:30 Jeff Kramer
Up adjust connections and basically edit the call flow in a flow chart which then was processed and ended up being the script on the system, which is very similar to what we have here, and you mentioned the maintenance aspect. They still required a lot of developers but moving it to the flow chart allowed people that were not developers to actually.
00:03:50 Jeff Kramer
Me be involved in the process and it was people that have a different skill set than the developers. These people were not necessarily that not that they had the engineering discipline, but they knew their product and they knew their product really well and they were able to make tweaks and actually update the product instead of actually having.
00:04:11 Jeff Kramer
To translate it to a developer that may not have the same domain knowledge, sure and that.
00:04:17 Spencer Reuben
Yeah, yeah. And I, I actually from a from the actual testing background, you know you used to have to go through each and every you know workflow right and actually had to figure out the workflow.
00:04:30 Spencer Reuben
Right. And you would have to go through each and every response and each and every decision point. And you know it was all manual at one time, even when you had the automation tools, we still had to get everything in there. And now she we’re reviewing what we’re doing with this with this knowledge graph is it’s amazing like because you can get.
00:04:50 Spencer Reuben
Full coverage, right and you get not only get full coverage, but you could also adjust it to get partial uh and set up specific workflows that you want to just test.
00:05:00 Jeff Kramer
Never the case that the tests were something that a non developer or non tester could understand. It wasn’t like you could turn around. I mean at least I’m assuming this that you could take these test cases and just basically present them as is to a third party say that’s in another department of the company and have them go. I can see what you.
00:05:19 Jeff Kramer
Testing this makes sense to me, or I’m assuming that’s the case, right? You didn’t.
00:05:23 Spencer Reuben
Have that exactly a lot of times we would actually have to a lot of phone calls, a lot of meetings to actually understand the workflows. And now with doing this with the knowledge base.
00:05:39 Spencer Reuben
It’s so much easier.
00:05:40 Jeff Kramer
Yeah, it was. It was the same thing there. It’s like they had. They were a big company. They had, they had the, the voice systems, but they also had to have a call centre for when people were using the voice system and needed.
00:05:52 Jeff Kramer
Help and putting the information in this flow chart which basically was a knowledge graph, allowed them to basically have one single source of truth and you could take that flow chart and give it to somebody that was in their call centre and customer support and show it to them and they would instantly understand.
00:06:12 Jeff Kramer
The flow and then be able to work and talk with the customer about it. You didn’t necessarily have to have all the documentation and all the other stuff that you’d have to do with a normal program. It would be that one thing was both the program and the documentation.
00:06:26 Jeff Kramer
And it could go to the marketing team, the support team, everybody basically could just use that and that to me is really impressive. Funnily enough, at the time, I didn’t think it was that impressive.
00:06:37 Spencer Reuben
Yeah, I agree 100%.
00:06:41 Jeff Kramer
At the time, as a developer, I’m thinking it’s much quicker for me to program this in another language than in a.
00:06:46 Jeff Kramer
Code chart and you didn’t necessarily see the benefit of having non developers be involved in this process. It seemed like the developers thought they were the actual experts on this and coming here I can see that although you still have, you still need some you need domain knowledge, but you don’t necessarily need programming knowledge.
00:07:06 Jeff Kramer
When you’re using the knowledge graph, and I think that the trade off I think is definitely in the favor of the company to be having a knowledge graph used to expose this information, then have it sort of be.
00:07:21 Jeff Kramer
In the hands of developers who are focused and specialized.
00:07:26 Spencer Reuben
Yeah, you know what the what? I see. There are really three main things, right. One is you’re getting the coverage right. You’re getting all the coverage of, you know, the entire workflow that is under test. Right. You’re getting the of speed that you’re getting the.
00:07:41 Spencer Reuben
Coverage and how long it takes to get all those tests up and being able to have them generated and executed right and then you have the flexibility, right, the flexibility actually allows you to change and modify this, you know quickly and easily. You don’t need to be a specialist.
00:08:02 Spencer Reuben
And as soon as you do that, then you have, you know, so many different ways to apply this testing.
00:08:07 Jeff Kramer
Yeah, I agree that that especially the not being having to bring not having to bring the specialist in is really it. The turn around time it’s much quicker when you do that the.
00:08:21 Jeff Kramer
Not having to bring in something to change the documentation being allowed to have basically anybody who knows the product being able to look at this and at a glance be able to make suggestions and you changing basically instantaneously based on that feedback is I think it adds a lot. It’s spreading the knowledge through a knowledge graph.
00:08:41 Jeff Kramer
Thing is, lots of benefits and then like I said, it was a little bit surprising when I started because in my first experience.
00:08:48 Jeff Kramer
I didn’t understand that, but now it does make total sense. It’s sort of funny. We never called it a knowledge graph back then, but that’s basically what these things were.
00:08:58 Spencer Reuben
Yeah. Anyways, Jeff, I got to get going. I got another meeting to run to but hope to talk about this again.
00:09:05 Jeff Kramer
Yeah, yeah, this is. You know, so much more fun. We should be doing this a lot more.
00:09:09 Spencer Reuben
I agree. OK, bye for now.
00:09:10 Jeff Kramer
Talk to you later.