TigerBalm part 1
===


### General Questions ###


##### GQ1 #####


investigator: Okay, recording in progress. So, now we continue your questionnaire with the general questions regarding usage of digital libraries. The first question is which tasks do usually use a digital library for, please tick all answers, which apply and complete your own tasks if they are not given in the pre-selected answering options.

Please give short oral examples of the tasks you're ticking. So, for example, you could say, I tick person search because I like to keep track of myself. So, what are the tasks you usually use a digital library for? 

TigerBalm: I'm ticking get full papers, full text papers. Well, because it's important to know like what the authors have written. Usually, I'm doing a concrete or precise paper search looking for latest papers especially literature reviews, summaries, and survey papers. They're very good if you would like to go into a new topic.

Seldom I'm keeping track of myself. yeah, usually I'm getting BibTeX data because I need to reference those papers. Like in my slides, or like, let's say in a repo or something.

And I'm studying relations. Yeah, absolutely. Because I'm usually using connected papers. That's an online resource. And so, this is very nice to study co-author relationships and to see like, who has referenced whom. 

investigator: Anything else that's not written here? So, you can also include your own tasks.

TigerBalm: Yeah. Maybe like on "Sonstiges", I would say using connectedpapers.com I'm not sure whether it's written with a dash or not. 

investigator: That's okay. I'll find out. 

TigerBalm: Okay, second question, or?

investigator: Oh yeah. If you think you have written down all the tasks.


##### GQ2 #####


TigerBalm: Which system doesn't, where do you oh, okay.

investigator: Okay. Yeah. Second task. second question is which system or digital library do you usually use to solve these tasks? Please tick all answers, which apply and name others, which also apply, but have not been written down here and please give a short oral description, why you like, or why you use the system?

TigerBalm: As I said, I'm starting often, very often with connected papers. then if it is also possible yeah, I'm using Google's scholar quite often. because what I like with Google's scholar is that yeah, it finds also like papers, which are not that high in ranking actually.

But yeah. How should I say that? It's not like just like yeah, listing up papers, which have like a high rank, but are more like more like a report or something like that. So, at least that's my impression that I find those contributions much more likely in Google scholar than in Scopus, for instance, because that's also another resource that I'm using quite often.

DBLP is something. In Semantic Scholar is something that I'm using quite often for for getting better cites. So, like I said, like before, like when I'm using like BibTeX data or something like that, and I'm trying to cite a paper, then I found out that personally, so my impression is that in Semantic scholar it's much better to just copy and paste those things then because you have different forms that you can select.

And this is something very nice and Springer Link is if I'm using or if I'm looking for an entire book on a certain topic then Springer link is also something that I'm trying to address immediately, or it actually points out that Springer link has a lot of those things. Research gate is something that I'm not using very often and using like a standard universal search engine, such as yeah, a Google.

Yeah. It's usually also not as I said, like Google scholar is fine, but using, let's say the conventional search slit in in Google is not something that I'm targeting it. 

investigator: Yes. That's nice. So, do you wanna add anything else to "Sonstiges" or is that it?

TigerBalm: Maybe ACM is also something like, depending on the topic, ACM is also something if it is more like an IT topic, I usually have the impression that the ACM digital library is also good resource. Yeah. Maybe this is also something that I would dash, but for the others I'm not so much experienced.

Oh, semantics scholar I have to dash then, sorry. Going next or? 


### TASK 1 ###


investigator: Yep. If you're finished with that then yeah. Let's go to the next page. Now you should see task one. Consider the following task, find two experts on a topic of your liking. Example topics could be "domain specific query languages" or "hashing functions" but should be from the general or broader area of computer and information science.


##### TASK 1.1 #####


You can pick any topic that you like, and you do not really have to do this task right now, but we are going to talk about how you would solve this task. So, my first question would be what is your chosen topic? 

TigerBalm: "Domain specific query languages". 

investigator: Okay. But you can use anything you like.

TigerBalm: Ah, any, anything you like?

investigator: Yes, yes, yes. 

TigerBalm: Okay. Then we do "domain adaptation". 

investigator: Okay. 

TigerBalm: Something I have to enter there or?


##### TASK 1.2 #####


investigator: No. You don't have to. Okay, perfect. And question two is how familiar are you with this topic? 

TigerBalm: I'm let's say starting.


##### TASK 1.3 #####


investigator: Okay. Great. and how would you define an expert?

TigerBalm: Since I found out that this is quite a new topic, let's say starting in 2018, 2019, I would define an expert, which is which contributed, let's say in the past three to four years, at least two or three publications, in this topic. and since I'm not only interested, let's say in "domain adaptation" in general, but let's say but with a certain application field, you know, I would consider an expert being a person which actually had the, in the best case, you know, like exactly something published in this application area. 

investigator: Okay. That's great. Do you want to add anything else to your definition of an expert or should we continue to the next question? 

TigerBalm: Since "domain adaptation" is very much related to let's say also to "generative models", you know so it's a machine learning topic and this is why I would most probably also broaden the definition of an expert also as somebody who's got, yeah, contributions like in this area as well, you know, like not machine learning in general, but let's saying, okay. let's say "generative models", something like that. Anything autoencoders and all those things. 

investigator: Okay, great. So, are you now happy with your definition or do you want to extend it or specify something else?

TigerBalm: No, I'm happy.


##### TASK 1.4 #####


investigator: Okay. Perfect. So, then let's continue to question four. How would you solve this task? How would you find two experts on a topic of "domain adaptation"? 

TigerBalm: So, what I, because I recently did that...

investigator: That's great! 

TigerBalm: And what I did was I was looking on connected papers. For let's say a paper title that I could imagine that was like "variational autoencoders for domain adaptation" which was not including yet the application area that I mentioned before, but that was my starting point.

And then I looked like for publications, because he was building up the graph. And then I felt like I was, or then I was first focusing on the bubbles that are larger than others, because they're much more referenced. And then I checked whether they were most recent. Or they were already let's say the published papers were at least let's say two or three years old.

Because then I felt like it was like somebody who started that research already, like a couple of years ago. And it's not, let's say somebody who recently published something in this area. Cause that was con referring to my definition of being an expert. Right. So, and then... Basically look from there. Like who's like publishing what, and then I noted down some names of the publishers of the experts of the potential experts, because I didn't note that by then. and then I looked them up on the web, like which institute they're from what else they're publishing and Then I found out that one of these guys was like coming from a certain application area, which I was not interested in. So, I skipped that one. another one was like from a mathematical Institute, which was just, let's say generally interested in that topic. Most just from a mathematical point of view.

but after I did that a couple of times, I found a person, which is exactly like contributing in the area that I was interested in which is basically "crop detection" and yeah, and this is somebody that I noted then and yeah, that was the first one I had had.

And you said like the second one, right? So, there was probably I would see whether these two experts have. So, I haven't done that yet. Right. So, but potentially I would check whether these guys have published something at the same conferences, or they have worked even together whether they're from the same country.

Or from the same, let's say continent or Europe or north America or something like that. And whether there are some relationships between those two persons and yeah, that would be in theory, the things that I would probably do. 

investigator: Okay, great. So, you mentioned quite in the beginning of your explanation that you would look at the authors how would you look at the authors, which authors of the papers would you look at and then to which system or digital library or whatever would you switch?

TigerBalm: Usually I'm switching to the libraries that are linked from connected papers because this is usually like a service they offer and the authors that I'm looking at usually are the ones that appear first on the head of the paper or as in the list of the authors.

And even knowing that this is sometimes not the best way to do it, because of many reasons. but anyway, so this is usually like what I'm doing to identify, but basically what I've done is I've been parsing the entire list of the authors.

So not only the first one, but I've started actually with the first one. Yeah. 

investigator: Okay, great. So, you also mentioned that you would check if authors or people have published at the same conference. Where would you find out about that and how would you do it? 

TigerBalm: When I'm checking actually the publication list of all of these authors, if these are authors from the academic field. And that's usually not a problem because they have usually like a publication list and then you can see like where things have been published. if these are authors from companies when it's like, let's say Google or apple or something like that or research Institute then it's sometimes I find, at least it's my impression, you know, that sometimes it's a bit more difficult to find out like what else they have published. And this is how I'm doing it. So, I'm basically going through the list of publications and then checking whether there are some same conferences or whether there's a conference that repeats all of the time. So, annually or like in two years frequency. 

investigator: Okay. And you also mentioned that you would check if different persons came from the same continent, where would you find this information? And yeah?

TigerBalm: Also, like let's say on the top of the paper. And then as I said, like, I'm checking also like the Institute they're coming from, because like, usually if there's somehow from the same Institute, let's say "institute of information science", then at least sometimes that works that there are more people like working for the same Institute contributing in the same or similar areas. Right. So, especially when it is comes to as I said, like I'm looking for certain, application areas such as "remote sensing" or "crop detection" then these are usually like our contributors from let's say agricultural Institutes. Since I'm not an expert in this area. I'm this is like, let's say the little "Strohhalm", the little straw that I'm trying to pick, you know, in order to find through the jungle of this domain, you know, because since I'm not an expert, so what else, what other chance do I have to access this? You know? 

investigator: Yes. Okay. and you also mentioned that you would check the relationships between authors. 

TigerBalm: Yeah. 

investigator: Would you use a specific system for doing this or?

TigerBalm: Yeah. As a, no, I'm not using a specific system, I'm just looking like when I'm seeing like they're from the same Institute or they're from the same let's say Fraunhofer or Max Planck or something like that, or like a research association just the first indicator and then seeing like the conference staff were published and so on.

So, if I'm really looking for, for somebody who is an expert. yeah, then I would do this manually. Okay. Or from an international point of view. Yeah. 

investigator: Okay. And you also mentioned that you would check the publication list of authors, which could be kind of difficult sometimes. Do you check the publication lists of authors in the digital library or to system, or do you check their personal homepages?

TigerBalm: Their personal homepages.

investigator: Okay. Yeah, okay. Perfect. 

TigerBalm: Always, to be honest. 

investigator: Okay. 

TigerBalm: Or on Google scholar sometimes. Yeah. Google scholar sometimes also, but very often on their personal homepages. 

investigator: Great. So, do you want to add anything else to this task?

TigerBalm: No, not if I'm looking for an expert. No. Okay. 


### TASK 2 ###


investigator: Great. Well, as that's the question, then we can continue to the next page to task two. Where we have another task consider the following task, find relevant papers from a topic of your liking, which appeared after 2017 example topics could be "paper recommendation" or "author disambiguation", but should be from the broader area of computer or information science. Here again, you can choose whatever you want, so you can stick to your previous topic, "domain adaptation", or choose a different topic. And yeah. 


##### TASK 2.1 #####


What would be your chosen topic for this task? 

TigerBalm: Right. So, the chosen topic for this task would be "crop detection" or especially like "coffee detection", "coffee, plant detection".

investigator: I'm not sure if this is really computer/information science-y enough. 

TigerBalm: Well, it is. Okay, well, actually it's a kind of a computer vision problem.

investigator: Okay. 

TigerBalm: Well then "remote sensing data". Right. So, yeah, exactly. So you have satellite images and you're trying to detect like, what is growing there? You're especially interested in coffee. 


##### TASK 2.2 #####


investigator: Okay. No, then that sounds perfect. thanks for explaining this. So, question two is how familiar are you with this topic? 

TigerBalm: Not at all familiar. Okay. 


##### TASK 2.3 #####


investigator: Okay. Great. So, then onto our next question, how would you define relevancy?

TigerBalm: Very difficult. So, relevancy here is it, so. Well, maybe I should start by what I, what were my problems when I started searching this this stuff. my motivation, first of all, is that I have currently a master's student from (ANONYMISED INSTITUTE) is another Institute here at (ANONYMISED AFFILIATION) and these guys or she is like writing her master thesis, in this, and I'm trying to supervise her in the aspect of yeah deep learning methods and so on. And so, the computer vision part. Right? So, and the point is that coffee seems to be a really difficult plant to detect, which I didn't know before has certain reasons which we go, don't go to into detail now, right now.

But the problem is that a coffee detection as you're already most probably have imagined, when I first said like coffee detection is really like a broad area, so you can do this on a plant basis, you can do like, let's say soil detection and so on and so on.

But you can also do this computer vision part and the computer vision part would include that you have remote sensing data that you have different bands. You have radar data. You have let's say vegetation indices coded in those images. Right? So, I'm telling you this because what I started with was like keywords. Keywords, like "remote sensing" "soil moisture" "vegetation indices", "remote detection", "cropping detection", and so on and so on.

And this is how I started with connected papers again. was trying to find papers, which also did that with deep neural networks and not with shadow learning algorithms, such as I dunno, random forests, which is very, very distributed, and very often used in this area. And well, this is how I started connected papers.

And then I looked again like what kind of papers are there, how often do they reference each other? And yeah. It's actually not too far away to say like which appeared after 2017, because we were interested in papers, which were focusing after 2018, to be honest, you know?

And so, from that point of view, that is similar. Yeah. So, relevancy in this respect for me was just like, let's say like a many, many, many, many steps, but first of all, let's say covering as many keywords as possible, which I defined and learned over time. To be, let's say best representing my research question. And then yeah, then going, just actually to connect the papers, which was my first touch point, always. 


##### TASK 2.4 #####


investigator: Okay. Yeah. I think I get a general idea, what's relevancy for you for this topic. So, now I think we can go to question four of you describing how to solve this task. Because you already kind of started with the description, but yeah. 

TigerBalm: Yeah. since for me let's say a relevant paper is also something which comes potentially from an expert, which it's not always the case, obviously, you know, but in, and very often, or in very many cases, it's the, this is applicable.

So, I would either go there and say, okay, let's try to find an expert and see what this person is then publishing or. I would say, okay. Also, recent publications, you know, by let's say a newcomer is also let's say relevant. So, that means like solving this task would be including all the steps that I've mentioned before going to connected papers being like, what are the relevant conferences for these ones? if I haven't done that before in the expert search, I would do it at least right now. And then see like what conferences usually cover papers or what journals, you know, not only conferences, but what journals cover, usually let's say, or publish papers which are containing the keywords that I'm looking for.

investigator: Okay. So, in which order would you do it? Would you first find an expert, or would you first try to find someone who's kind of like a newcomer for this area? 

TigerBalm: I try to find an expert, to be honest. 

investigator: Okay. And when would you decide on finding a newcomer? 

TigerBalm: When I'm more familiar with the topic.

investigator: Okay.

TigerBalm: Because this is really like my biggest issue. You know, I'm very often researching in areas, which I'm not myself, an expert and this is why I don't know these people so very well. And this is why it's very often too difficult for me to access this information or to evaluate this information.

Because if I would be myself an expert, then I could immediately know why, what this, this and this person publishes is always good. And this person must be a newcomer because I don't know this person. Because I can't do that then. yeah, that would be the first thing to do for me.

investigator: Okay. you mentioned before that you start with like one keyword and then you kind of expand the keywords by terms that you find, how exactly does that happen? 

TigerBalm: Yeah, I, most of the time I discriminate let's say the search down to very few keywords and then just orient myself and then see, okay, what else is there? And then I found out, okay, coffee detection is really like a broad field can have to do with agricultural aspects as well as with computer vision aspects. So, I took like the computer vision keywords in there. and then since I was only interested in, let's say, deep learning algorithms, rather than shallow learning algorithms, then I took these ones in there and this is how I'm proceeding.

investigator: Okay. 

TigerBalm: And well, I'm discriminating. I'm sorry. So, I'm discriminating, let's say my search amount of hits and I'm proceeding in this, you know, so I'm very broad in the beginning and I'm discriminating more and more. 

investigator: Okay. and you do these searches on the connected papers?

TigerBalm: Most of the time I'm doing connected papers. But as I said, like before, like on the first questions, also, I'm sometimes I'm using Google scholar or like on yeah, Scopus and so on. 

investigator: Okay. are there any specific decision criteria when you switch the platform? 

TigerBalm: Not really. No. 

investigator: Okay, then I think you answered all my questions, but do you want to add in something to this task?

TigerBalm: Yeah. Sometimes it's And I mean, I'm not sure whether this is interesting to you guys, but within your study, but I'm very often talking also with people from other institutions or something like that. And then sometimes asking like what they could recommend me to get into this topic.

Right. So, in this special case there was like a master's thesis that I'm supervising together with the (ANONYMISED INSTITUTE). And the (ANONYMISED INSTITUTE) has another research scope. And there you have other experts. And when I said, like to, "Hey, what paper should I read if I'm trying to get into coffee detection?" and then get some recommendations. That's also something that I'll sometimes do. 

investigator: Would you do this at the beginning of your search or after some time?

TigerBalm: Yeah, after some time after I've at least a slight idea of what I'm looking for. 

investigator: Okay, great. And then you get back to searching with the papers or the keywords that have been provided you. 

TigerBalm: Yeah. 

investigator: Yeah. Okay. Good. Anything else you want to add to this task? 

TigerBalm: No, I think this is it. 


### Thank you ###


investigator: Okay. Well then thank you so much and I can stop recording.
