Episode 17 - Eirini Komninou, data science and sonification
Transcript
Graham Johnston and Kirsty McIntosh talk to STA volunteer Eirini Komninou
Graham Johnston 0:00
Hello and welcome to episode number 17 of the Scottish Tech Army podcast. I'm Graham Johnston
Kirsty McIntosh 0:07
and I'm Kirsty McIntosh. Welcome back, Graham. I'm glad to find you again.
Graham Johnston 0:11
I know, I've been busy preparing for lockdown 2.0 in my day job, so yeah, it's been pretty all encompassing in the last couple of weeks, but yeah, I've missed it.
Kirsty McIntosh 0:23
Of course you have, of course. Well, welcome back, anyway. Today we have the fabulous Eirini Komninou, who is one of our Scottish Tech Army volunteers, and has been very proactive on our homebrew challenge COVID-19 dashboard for Scotland. So, welcome to the podcast. Eirini.
Eirini Komninou 0:42
Thank you. It's great to be here.
Kirsty McIntosh 0:44
Can you tell us a little bit about, I suppose first how you arrived in the Tech Army? What brought you to us?
Eirini Komninou 0:49
Yeah, sure. So it was right about the time that I was finishing writing up my PhD thesis. And it was right about the time that we started, you know, things started closing down because of lockdown, number one. Therefore, I thought, I don't know how to help my local community. I come from a computing background and so I thought the least I can do is just try and do some so called number crunching. And around the time someone told me that, by the way, have you checked the Scottish Tech Army? So I volunteered for the Scottish Tech Army hoping that I can apply some number crunching.
Kirsty McIntosh 1:35
So if you've been here for what, three or four months, then sort of, right from the very beginning?
Eirini Komninou 1:42
It was from late April.
Graham Johnston 1:45
That's pretty much the start no? I mean, it was March wasn't it, end of April?
Kirsty McIntosh 1:49
No, no it was the end of April, actually. It'll be interesting, Eirini, you may be a first responder. Maybe one of the first people that signed up, I signed up on day one as well. I got a little badge that told me I was a first responder. I'm very proud of it.
Graham Johnston 2:05
Yeah, exactly. I'm quite jealous of that actually, I've said that on some previous episodes, you know, I missed that first responder badge. I didn't get it. But I was soon after responder. And Eirini, really, we can't go any further without you telling us what your thesis was in. I'm fascinated about this. What did you do your PhD thesis in?
Eirini Komninou 2:25
Yeah, well, if you want, if you if you have trouble of sleeping at night, then I can tell you a bit about my thesis , it's perfect for insomnia. There is a subset of algorithms called meta heuristics. And they can be applied for optimising scheduling, among others, or route planning and so on. And so I took these algorithms, and they applied them for scheduling satellite operations. So it was quite an applied project.
Graham Johnston 2:55
That is unbelievable. That's, that's not boring in any way. I mean we could probably do two more episodes on this.
Kirsty McIntosh 3:04
You'll coming back, and we'll be doing a series with Eirini Komninou! So tell us about how you got into data science in the first place then.
Eirini Komninou 3:17
So since my PhD was on this small subset of algorithms, data science only started becoming a thing when I was kind of halfway through my PhD journey. Therefore, I had to learn how to become a data scientist by doing. And that meant looking up a number of data sources of interest. I'm interested in the medical applications of data science, and then trying to make inferences based on data. This is what triggered my interest, if you will. And then I was lucky when I joined the Scottish Tech Army, I realised when I joined the dashboard team, I realised that there's plenty of data to be processed. So I guess that was a good opportunity for me to brush up on data science.
Graham Johnston 4:17
And so for those listening, the dashboard that Eirini has talked about there is the COVID-19 dashboard that the Scottish Tech Army produced, which is visible on our website, which is scottishtecharmy.org. And it's under the About section as COVID-19 dashboard. So really, what was it like when you first got involved in that project?
Eirini Komninou 4:41
It was daunting to be very honest, because we were handed over we were given access to open data from Public Health Scotland, and there's quite a lot of it, I must admit, and our job was to try and make it more digestible, if you will for those who are not particularly data science savvy. So obviously, we have a responsibility as well not to misinterpret the data and not to spread panic, and things like that. And then we had a good head scratch as to how can we present the data as well as possible without distorting the message?
Graham Johnston 5:27
Well, I can tell you from someone that is an avid fan of, of what you've achieved, and looking at that dashboard on a regular basis, it is very digestible. I think that the way that the data is presented is, is amazing, actually, and everybody I've everybody I've talked to, and everybody I have introduced to the dashboard has been absolutely blown away with just how well presented it is and how agile it is, and how flexible it is in terms of being able to look at all of the different aspects of it, because it really does, I think, help a lot, certainly, people I've spoken to it helps a lot to understand the impact and the different areas and look at all the different measures, and it's constantly being updated and revised and improved as well, which is which is fantastic, too.
Kirsty McIntosh 6:20
Yeah. I concur. Yeah, it's absolutely fantastic. It tells a grim story, but it tells a really interesting and useful one as well. So you're all to be congratulated on it. What's it like working in that busy team, there's an awful lot of people on that on that dashboard team - must be fun from time to time?
Eirini Komninou 6:40
It is good fun. I've met a great deal. Very, very interesting people very, very helpful people, generous as well - generous with their time, generous with their knowledge. So yeah, I'm impressed. People are passionate as well. So I should note here that this is teamwork. The dashboard that you see is teamwork. It's a good few people that have worked very hard for it. So thank you for your kind feedback. And kudos goes to the entire team. I should note here that the entire team are very passionate, like you see people working on the dashboard after their day job, despite having families to look after and so on. It's really fascinating.
Graham Johnston 7:31
Yeah, it's an incredible commitment, isn't it really, from everybody involved, all volunteers. People putting in their day job and then doing it afterwards as well. It's just a great commitment. And I think everybody's so bought into the project and the value that it brings to people who do use the dashboard and the information that it that it produces that it's really rallied people, which is phenomenal. Yeah.
Kirsty McIntosh 7:57
And it's also been a brilliant opportunity for foreign volunteers that have come in to the Tech Army to learn some new skills and, and sort of brush up perhaps on some old ones as well, hasn't it, Eirini?
Eirini Komninou 8:09
Yeah, absolutely. It has been a very eye opening experience, especially for those of us who just finished our degrees or, you know, Codeclan, and similar bootcamp graduates. It's teaching us how to work as a team, how to collaborate efficiently and effectively.
Kirsty McIntosh 8:29
So you have suffered from visual impairment and this sort of led you to do something pretty spectacular on that dashboard. Tell us a little bit about the story. You were present on the day that they presented to Scottish Government and heard somebody say something?
Eirini Komninou 8:44
Yeah, certainly. So a few years ago, because of illness, I lost my eyesight. And slowly over the next couple of years, maybe I regained the good part of it. But having tasted what it's like to have non functional vision, I greatly appreciated the importance of accessibility. Now, fast forward in July 2020, when our team presented to the Scottish Government to very knowledgeable officials, who are dealing with the data generation and presentation in the Scottish Government, during the Q&A session one question we received was, how do you tackle accessibility? How accessible is your dashboard? And how are you going to improve on your accessibility? And then that person who is by the way, the Head of Service Design of the Scottish Government, with a background in human computer interaction in computer science, so that person said what about sonification Have you considered that and I kept that name, that word rather, term because not long before this presentation, I had watched a presentation by a blind astronomer. That blind astronomer did her PhD on the topic of sonification. Because that's how she can do her job, by listening to data trends, and I found it fascinating and I found it fantastic and I found it a very forward thinking thing to do in science. And that's how it all started. So I approached the team, our name is DataSTAr and they embraced the idea, we started working on it, we prototyped something that can turn Public Health Scotland data into sound. And then we took it from there and put it into our production dashboard.
Kirsty McIntosh 10:59
And it's absolutely incredible. And you know, even as a sighted person to listen to that data isn't it Graham?
Graham Johnston 11:07
It's totally amazing. We had it on a few episodes ago. And I think we actually played the sonification over the top when Alistair was talking about it. And the reaction it's ha from people is amazing. I mean, the first time I heard about it as a concept. I thought, this sounds absolutely amazing, I need to experience it myself and I've shared it with multiple other people, and it is absolutely the standout that they talk about is the way that that brings to life. You know, as you said earlier, Kirsty is a grim set of statistics, but it brings that to life. In the same if not more impactful way, than the visualisation does within the dashboard. So absolutely, hats off to you. It's an amazing advancement. And I think as per the press that has picked it up, just shows how forward thinking it actually is and kudos to you for, you know, bringing the idea into the team and making it happen in such a short space of time. And I'm really interested in the process, Eirini if you wouldn't mind, in terms of just how you do apply that sonification to data. Is there a particular platform plugin, is there a programme that does it, automatically how much work is involved in actually turning data into sonification?
Eirini Komninou 12:34
I'm not the expert on the dashboard side of things, the certification side of things, because we had our whizz kids, JavaScript experts who turned data into sound. I can tell you what I did in the prototype stage, which is what I coded. But generally, what we have done is to take each number in our data - it's a time series data. So it's data over time, from the 28th of February, until, for instance, today or yesterday. And so we take these numbers, which represent Edinburgh, for example, the council area. And then we turn this into audible sound into something that we can hear, which is I don't know much about MIDI (Musical Instrument Digital Interface), but I think that MIDI is a standard used for producing digital sound. So what we know is that audible sounds for MIDI are between zero and 127. So we need to take whichever number, for instance, daily cases, the number for Edinburgh, let's just say and turn it into a value between zero and 127, which is our normalisation of data. And then we would take that number and pretty much work as low in terms of hardware as low as possible so that we can force the computer's soundcard to produce that sound, if that makes sense.
Kirsty McIntosh 14:17
I was just saying, it's not exactly preheat the oven, you know, bake for 20 minutes kind of stuff is it? It's sounds really fascinating. I think also there's something really inspirational about this kind of thing. You know, I think I've said this before on the podcast that you know if you can show children that they can effectively, you know, data, you can make music from data you can you can do anything with data really and actually, it's an exciting thing to learn and know how to do I think, you know, examples like the sonification of the dashboard are brilliant things to sort of show and inspire children as well. I think it's great.
Graham Johnston 14:56
Sure is.
Eirini Komninou 15:01
Kirsty, I've heard you saying before, about talking about the value of showing children, how we can turn data into sound. And I totally agree with you, I think you're very forward thinking. Not only is it inspirational, I would like to think, but also, one thing that you've picked up on is that sonification is also used for augmenting data flows. So even fully sighted people can actually use sound that will augment their visual perception of data. And that will help them pick up on less pronounced effects described by the data.
Kirsty McIntosh 15:43
Yes, it's almost like a more instantaneous storytelling there, so the highest pitch sound tells you it's really bad right now. And actually, you hear it so much more dramatically than you see it. So that's very true.
Graham Johnston 15:56
Yeah, I can, I can completely and utterly agree with that, because as soon as you shared it with me, Kirsty, I sat down with my nine year old and played it, and he was blown away. And actually, on a few occasions when we've checked the dashboards, and obviously has the sonification built into it, he's actually following it with his finger. So he hears a sound so he follows it with his finger along the graph, so that you can sort of see exactly what it is, as you say, it absolutely does augment it, and it makes it...it's kind of like that sort of full 4D type of experience, isn't it? So you're immersing yourself in the data, and you're seeing it, you're hearing it if you're a sighted person and being able to, you know, follow it through. So I totally agree with you. It's, it's an amazing advancement.
Kirsty McIntosh 16:46
I think we need to definitely put the call out there for some of the organisations we've been helping if you're sitting on a silent data lake and you need it, you need it sonified and visualised, then get in touch with Scottish Tech Army. I'm sure we'd be happy to help. We're definitely up to the challenge. I certainly think that's the case what with 1200 volunteers to draw from, which is fantastic. Eirini, you've just started a new job as well. I just wanted to say congratulations to you on that. It's always nice to hear a volunteer, going back out into the workforce, which is super,
Eirini Komninou 17:19
Thank you. It was through the STA that actually connected with that company. So thanks very much for that.
Graham Johnston 17:26
Oh, that's fantastic news. Congratulations, brilliant.
Kirsty McIntosh 17:31
We can take the credit, excellent.
Eirini Komninou 17:36
You should take the credit indeed.
Kirsty McIntosh 17:38
Not at all, it's always great. It's really good to hear it's, it's part of the journey of the volunteers. You know, we want our volunteers to be working. And that's the biggest success for us as well, which is which is fabulous.
Graham Johnston 17:51
Absolutely. Can I can I just sort of segue slightly Eirini and just try and take sonification as a topic. How much sonification is actually out there, because it was the first time I heard about this was through the Scottish Tech Army that sonification of data and realising that it was so impactful. Is there much sonification out there? I mean, I think I know the answer to the question of the second question, which is there enough out there? Clearly, there's not but are there other examples of sonification that you've heard and can share with us?
Eirini Komninou 18:34
From the background literature review, if you like that we did in order to understand what's out there in terms of sonification we found that as I mentioned earlier, astronomers are perhaps the primary users in terms of science. To my understanding, they use sonification quite a lot in high energy astrophysics, for example, X ray astronomy, which that astronomer that I mentioned earlier, who did her PhD in sonification, she's a high energy astrophysicist, for example. Other than that we found a at least a few projects, art projects, where for example, someone was presenting the climate change, essentially global warming in the form of sound. And you can tell it's quite, would you call it instinctive, you can actually tell there the huge difference between for example 100 years ago, and today and how fast within the last 50 years, we saw the globe, warming, the earth warming up.
Kirsty McIntosh 19:53
I don't know I want to hear that, actually!
Eirini Komninou 19:57
It is grim, as you said but at the same time, it's also in my opinion very informative. Because it gives us instead of just reading, you know, instead of going through an arduous scientific articles or even visualisations, we can actually hear the result, which is obviously sped up. But still, it gives us a very good gut feeling of what's going on.
Kirsty McIntosh 20:25
Yeah, I was actually just thinking - this is probably gonna sound completely weird. But I was just thinking, if you could sonify the development of a baby before it's born, you know, that kind of, kind of every day, every week sort of thing as it as it develops and stuff like that. That might be for an interesting and interesting exercise as well, you know, something to study to give your kids when they turn 18 - here's a soundtrack of your development, I just, I suppose, you can practically use it anywhere actually, wherever there's data you could sonify it couldn't you? I think that's, you know, that's, that's the lesson taken from here.
Graham Johnston 21:07
100% 100%,
Kirsty McIntosh 21:11
So what's next for you then, Eirini?
Eirini Komninou 21:14
What's next for me, I will not join the reserves, I will remain an active part of the Scottish Tech Army. Because I'm proud of that. And I've learned a great deal. And I've made a good a good bit of, you know, quite a few friends in the process. I am quite passionate about accessibility. So in any way that I can help with accessibility, then yeah, count me in. And having said that, the more I go check Internet websites anyway, the more I check the Internet, the more I realise that it's not as accessible as it should be. In my opinion, we see nice and shiny websites and webpages, which, when you try to read them using a screen reader, they fail in so many different ways. So anyway, I'm digressing here. All I'm saying is that I am passionate about accessibility. I am still part of the Scottish Tech Army, and I'm here to stay.
Kirsty McIntosh 22:22
That's the right answer. Yeah, that's absolutely wonderful. And you're absolutely right, we really must be on the case of accessibility. I think, you know, we talk a lot about digital exclusion, digital inclusion, but you know, we've got to stop talking about it and start making sure that we're doing it as well. I think that's a good point that you make, and we should all be beating that drum really.
Graham Johnston 22:48
100% agree. Well, the Scottish Tech Army is a better place with you in it, Eirini, thank you very much indeed for joining us and sharing your story. And, you know, really inspiring, it just shows how you can take an idea, implemented, run with it and have such a massive impact on on many people's experience of understanding the data of something that is very important. And something that has certainly I think helped a lot of people to understand the trend and the severity of the situation. Incredible, incredible innovation. So thank you very much indeed.
Kirsty McIntosh 23:27
Yes, thank you, Eirini.
Eirini Komninou 23:28
Thank you very much. Thank you very much as well. You're being very kind and very supportive. I really appreciate that. But kudos goes to the entire team - DataSTA. They are the ones that embraced the idea and they came up with the implementation. So thumbs up to DataSTAr.
Kirsty McIntosh 23:50
Excellent. Hear hear to that.
Graham Johnston 23:53
Well done DataSTAr. Amazing.
Wendy Stubbs 23:55
Our podcast music is by Whisky Kiss. Our logo is by Helen Davis a part design. If you'd like to know more about what we do and how to get involved. You'll find us at www.scottishtecharmy.org or on LinkedIn and on Twitter at Scottish Tech Army