My name's Andrea Mihalik. I am with the co-founder of Plum Analytics. And right now, I work Plum, and I work for Elsevier. So in thinking about this session, what's new around altmetrics, what's

new around research metrics, time has moved on. So we began the term altmetrics was coined back in 2011. And so you think about the expansion of just even talking last session, the previous sessions, about how are these metrics used now.

They're becoming much more common, but we really need to see how they're used in practice and what's the new thinking here. And I don't, you know, has anyone seen my data? Has anyone seen Nick Sheppard? Because Nick is our first panelist,

and it looks as if Nick is not here. So it's kind of an unfortunate title for his talk at this point. He is here.

Yes, so I will keep talking. I can so do this. You know, as you're introducing a session, you really think I need to keep my opening remarks brief. And now this is an example of when I really shouldn't have kept my opening remarks brief. So the question about new uses of altmetrics, I think in our previous panel, what Patty was talking about

are alternative research outputs. And we definitely see a lot of movement about talking beyond the journal article. We also see a lot of movement in the space about talking before that journal article gets published. You know, what can happen with preprints?

What can happen to all the different versions together and pooling that in one place? So data, I don't want to steal Nick's thunder because I'm sure he'll be right back, is when we're looking at all of these different forms of research output, there really

is data exhaust that's out there about them. And often the challenge is even finding the people who are creating this work, these alternative outputs, the knowledge that they can come and that they can gather this sort of data. And there's value in doing it.

So there's value in collecting this for their impact case studies. I think Patty's examples were spot on about ways we can talk about things and how to do them. Sorry about that. No worries at all. I was so engaged in the conference, I completely lost track of time. So I just had to pop out for a moment.

Sorry about that. With no further ado. So hello, and sorry for the day. So my name's Nick Sheppard. I'm the Open Research Advisor at the University of Leeds. And I'm going to be talking about some fairly preliminary work I've been doing looking at our metrics and research data. So by research data, of course, we

mean all the various different types of outputs that we've been hearing Patty and others talk about, which increasingly, again, as we've heard, being managed as research outputs in their own right. And of course, data repositories are routinely allocating DOIs for that material.

So at the University of Leeds, for example, I'm part of a team that manages an institutional data repository. And we allocate DOIs to data sets through the British Library, which are routinely tracked by artmetric.com. I'm aware, obviously, I'm aware of our chairs

from analytics and there are other services out there. But that's the one that I've been using and the one I'm most familiar with for this work. So the reward mechanisms for, again, as we've heard, I think, for data are still somewhat immature. That data citation's limited. It's not easy to track for those technical reasons.

So I was interested in how metrics might offer a low barrier method to track engagement with data sets. So I'll show you some results from some UK-based institutional data repositories. Bit of a spoiler at the outset, the results aren't terribly, they're a little bit underwhelming. The numbers are low and the quality as well, to be honest.

So I'll say that at the outset with a little bit of preamble around the context. And with that in mind, we'll talk a little bit about the culture, again, as we've heard a little bit about the issues of actually sharing this type of material and how artmetrics can play a part in that and how data repository managers and other stakeholders involved with research data can encourage and facilitate

data sharing through social media networks, blogs, and Wikipedia and the various other platforms that these types of services track. So again, I'm sort of repeating what's already been said to some extent. But obviously, at a conference like this, I think the majority of people are thinking about traditional publications.

I'm sure people will be very aware as well of the, in the UK and European-wide, the emphasis on open access, particularly with the requirements for the next research excellence framework for all that was to be open access. And I've long been interested in the relationship between open access and open metrics.

I'm interested to hear what Mithra has to say on open access and impact. Because obviously, as again we've heard in terms of to tell the story, the data, the actual material has to be available for laypeople and the public to actually be able to access, you know, and not get behind a paywall. And again, as we've heard, the metrics published their top 100 on an annual basis.

At the same time, there's been a focus on open research data, driving activity in universities, certainly in the UK. So the Concorda on open research data is a seminal paper from 2016, and certainly driven a lot of our activity at the University of Leeds. That emphasizes that the primary purpose of research data is to provide the information necessary to support or validate research projects,

observations, findings, or outputs. And in addition to institutional repositories, there's a wide range of more general data repositories, such as Dryad, the UK Data Archives, and Eredome and Figshare are some of the more well-known ones, all of which allocate DOI's. At the same time as well, we're moving, I think,

towards a more cohesive view of open research. So people thinking about actually sharing the whole research process and the publications and the underlying data in a cohesive way. I particularly like the perspective of Wellcome as one of the funders that, they have the most clear, one of the most clear

explanations of what they expect from their funding in terms of when they explicitly refer to research outputs. You know, we're not just talking about the publications, we're talking about the full range of material, including data, software, and materials, and a few other funders are coming along those lines as well. Even the REF acknowledges open research now as well,

and it does indicate that there will be some sort of credit to open research and going above and beyond the open access mandate and activity to share research data. So these are some examples of data repositories. So the one in the top's our institutional repository. Zenodo here, so that's a big European funded

repository out of CERN, I think, which does actually have an altmetric donut in there. A lot of data repositories don't feature those as well. This is another one associated with the University of Leeds, I'll talk about this particular data in a little more detail in a moment. But the crucial point here, of course, is they all have adequate metadata to cite formally.

And certainly in terms of the concord that open research, it was made explicit that all users of research data must formally cite the data they use. However, there is not really currently a standardized method, certainly in terms of machine readability. So if I cite my own data in a paper, and also cite your data in a paper, there's no machine readable way to sort of

differentiate that, there's some work I'm aware of going around along those lines. So, and also many journals still include data as a supplementary information, you know, it's still not sort of formalized exactly how this material should be shared, and that data is not going to have a unique identifier, it's not going to have a DOI, it can be tracked down, et cetera, et cetera, it might be behind a paywall, it's not curated in the same way.

So certainly our advice at Leeds, and in a wide range of other universities, is that data files should be deposited and properly curated in a recognized repository with adequate metadata to enable citation. And there should be a data availability statement in the body of the paper, and as an entry in the reference list.

Quick word on peer review, again this is an area, and we all know how overworked and underpaid, or not paid at all, peer reviewers are. So I think there's a consensus that ideally data should be reviewed, but it's certainly not standard across the sector. There's one or two journals that are here with the Journal of the Royal Society Interface

indicating that, you know, they will, they may review data, but anecdotally certainly, you know, it's not something that happens regularly. And obviously we're talking about much more complex informational objects than the journal also potentially, could be hundreds of files, dynamic objects, all these kinds of issues. There's an interesting article from Top Carpet to link through there.

So to think again about citations, there's a couple of tools, there's several tools sort of emerging to try and track data citations. Probably the most well-established is the Data Citation Index from Clarivate Analytics, formerly World of Science. Limited coverage is a subscription-based product. We did have access at Leeds, which has now lapsed.

Another one worth exploring is Swell Explorer, which again a European-funded project looking at access to a graph of links between data sets and literature. But again, inevitably this misses citations, many data citations are informal, they're hard to track, and even where data are properly cited, not all publishers pass that information on properly, if indeed there is a way of passing it on properly.

So again, there's an argument that potentially, oh, metrics can, can it fill in this gap? So to return to that data set I mentioned before, this is, it's from 2009 actually in Data Dryad. Now the actual paper's very highly cited, nearly a thousand citations according to Scopus, but according to the Data Citation Index,

it's cited once by the original citing paper, so it's not even a proper citation. Looking at the Swell Explorer, I found one, evidence of one real citation in an elsewhere journal, which is this one here. And that's got a very clear citation in the reference that has been picked up Scull Explorer. The data set's been downloaded 30,000 times from a manual search of our repository at Leeds,

I found 25 citing papers where it's clearly cited. Dimensions from Digital Science is another product that's looking at this area, and I was able just by tracking the DOI string to find 198 hits there, which implies that the data is certainly been cited a lot more than indicated by the established tool.

So the results, no doubt, you've been working with data, as I say, my data source was Iris Data UK, so this is a beta service from GIS, which currently tracks 27 UK-based institutional data repositories to provide counter-compliant download statistics. I'm not quite sure what GIS plans for it are,

I think it's been in GIS for a little while, but it's an easy way for me to get DOIs, so I thought, from UK-based institutional repository. Now, as I say, the data wasn't especially exciting, there was only one with an altitude score above 100, only 15 with scores above 10, and mostly Twitter. There is a little bit of quality in this one,

you can see there's policy documents and Wikipedia and various things going on with the single highly cited paper from an academic at the University of Sheffield, but by and large, the numbers were underwhelming, to say the least. In addition, it's worth mentioning, several repositories aren't exposing DOIs properly, I was either finding publication DOIs or no DOI at all, without going into the technicalities,

I think this is associated with metadata scheme and the fact that, again, this is the cultural aspect that we need to work harder to make sure that data DOIs are shared with the same rigors as formal publications. There's the spreadsheet of the results again, it's just DOIs, et cetera, all these data's in the public domain so you can access it and look at it in more detail, if you're interested.

To return again to the data set that has certainly been selected more than once, as indicated there, were the altmetrics any better for that? Well, not really, it was 23 and you'll see at least one of them's me sort of asking about this data set on Twitter, so it's not especially exciting. Because the results were so underwhelming

from institutional repositories, I had a quick look, I managed to get a bunch of DOIs from Data Drive and UK Data Archive, I've run those against the altmetrics API. Again, there was very limited evidence of sharing and quite a few anomalies as well that I know altmetrics are looking at at the moment, so the numbers didn't seem to quite add up.

And the UK Data Archive, I don't think the way their DOIs were resolving wasn't actually trackable by altmetrics at all. I don't know if I've got time. Go for it, three minutes. So, family faces, you know, but just so differentially, Data Saves Lives was actually there when I captured the screenshot,

so there's a lot to be thought about here in terms of actually exploring this area further. I think we really need to promote a culture of data sharing and as Patsy earlier was saying, in terms of promoting different types of material, promote data as the scholarly output in its own right,

through repositories and DOIs, a move to open research, you know, make sure there's a culture showing publications and underlying data, encouraging engagement through social media, Twitter blogs, and Wikipedia. Wikipedia, I think, is a particularly interesting area that I'll just touch on laterally to talk about an award I managed to win accidentally.

So this was the first Data Management Engagement Award sponsored by Spark Europe, the University of Cambridge in GIS last year. So my proposal was to link RDM with the Open Science Movement through Wikimedia Suite with tools by essentially sharing openly licensed research materials with DOIs and repositories through Wikimedia Commons,

which can then be used to improve Wikipedia. I'm certainly on a bit of a learning curve because I'm not a Wikipedian by any means, so I've got a lot to learn in that regard. This is just a quick example of that. So obviously we know our metrics will pick up. We've got citations from Wikipedia, so this is one from our repository with a DOI

that I'll link back through to our repository that's uploaded to Wikimedia Commons and cited on Wikipedia and then actually picked up by our metric. So yeah, if you would like to know more about that project or help me out, as I could do with some, you know, especially if you're into Wikipedia, you know how it works, there's a form that you can look us to your interest

in some related blog posts. Okay, thank you. Okay, well, thank you very much. Thank you, everyone. Thank you also, the organizers, for placing this presentation. This is a specific session. I kind of like a lot this idea

of new thinking of our metrics. And somehow the work I want to present today that is a work in collaboration with my colleagues, Paul Bauters and Sara Saidi. It's a little bit of a reflection after several years of doing research about our metrics on how can we establish different perspectives

on how to use them, how to look at them, how to perhaps include them in some forms of research evaluation and perhaps even some forms of new research evaluation. Some of these reflections is what I would like to share with all of you today. So what I'm going to discuss today

is just a tiny portion of what we presented in a book chapter, the Springer Handbook of Science and Technology Indicators that you can find in the archive link I'm presenting here, the open access version of this book chapter. And what I want to present today is basically two main ideas.

On the first hand, a kind of dichotomy between what we could consider the more evaluative discussion about metrics, and on the other hand, the more evaluative or comparative side of our metrics. And the second perspective is how we can look at indicators in discussing the appropriation

and evaluation from a point of view of what is their scholarly orientation or scholarly focus or the social media focus. And these are going to be the two main ideas I would like to discuss with you today. So let's start then with this dichotomy between descriptive and comparative,

let's say social and media metric approaches. I would relate this to the more traditional bibliometric work. So in bibliometrics, there is a traditional distinction between descriptive versus evaluative bibliometrics. I would say also comparative bibliometrics. In the comparative approach,

typically the considerations that are there are related to identifying higher scores, lower scores, better, worse, highly ranked, lowly ranked, all these type of considerations. However, in bibliometrics, we also have a more descriptive perspective.

And with this descriptive perspective, we focus more on the who, so who are the authors, who is funding, who is collaborating, where are they doing the research, when is this research being done, how is done, what are the ideas that being considered in the development of these new research, and what is happening to this research, for example, in terms of citations.

I would say that in all metrics or in social media metrics, we can have the same approach. We can talk about some sort of comparative social media metrics. There have been some discussions about normalizing indicators, like field normalized mandatory readership, also some discussions about field normalized Twitter indicators.

There have been also proposals of some social media-based factors, the idea of the twin path factor, the T factor that is the HC index, but with tweets instead of citations. There are also some composite social media indicators, the RD score, but I would also say here the metric attention score, trying to combine different indicators,

which essentially when you combine different indicators, what you are stressing is not the descriptive side, it's actually the comparative side, because then what you want to do with that indicator is to compare units. And also some other more, well, comparative, evaluative, network-based indicators. My impression is that we haven't paid

too much attention to the more descriptive side of social media metrics. And here we can discuss indicators about coverage, about the density, about the presence of publications, outputs, datasets on social media, what are the trends, are the trends growing, are the trends decreasing? Another perspective in this descriptive social media metrics

could be the social media landscapes. So for example, which research topics are attracting more attention on Twitter, on Facebook? Publications from which countries are having a stronger presence, a stronger reception in these social media platforms? And also some more exploratory network approaches.

For example, the approach of analyzing the communities of attention around scientific publications, the analysis of Twitter coupling, so seeing how scientific topics are being connected by Twitter users, just to give an example. Or for example, hashtag coupling, how Twitters using or mentioning scientific publications

on Twitter are using different hashtags to refer to the same set of publications. I can't give examples for all this, I can just show a few of them, a few of this more descriptive side of social media metrics. I will start with the Twitter thematic landscape. In this case it's for the African,

for the publications producing the African continents from institutions from Africa. It's production in the web of science. This is a map of science based on topics. So it moves away from the more traditional classification and it creates clusters, well the nodes represent clusters of papers

that are connected by citation relations, meaning that every cluster represents a small topic of focused publications on that topic. The size of the node in this map represents the number of publications that come from African institutions, from African researchers. And the color indicates the proportion of publications that have received at least one tweet.

I have chosen here one tweet, at least one tweet. I could have chosen, for example, at least one tweet from a set of specific Twitter users that are relevant for me. Or for example, a number of tweets that is higher, both the average or both the median or any other indicator that I would consider. Here by selecting the proportion of publications

with at least one tweet, I'm basically looking at the coverage. So what is the coverage of African publications across the different topics on Twitter? The interesting thing is that I have highlighted a few of these topics, particularly focused on those that have a large activity, so the nodes are big, and also the proportion is high.

So red in this map indicates that at least half of the publications from African institutions are having some reception, some coverage on Twitter. So then we can spot topics like HIV, AIDS, sexual alternative diseases, maternal mortality,

as topics where African research is targeting a lot of substantial publications, and these publications themselves are attracting a relatively important share of attention on Twitter. Remember, this is the share of publications. By looking at the share of publications with at least one tweet, I'm somehow reducing the effect of, for example, outliers.

So one highly-tweeted paper that will somehow influence the whole analysis. But of course, we could use also many other indicators. We could use Facebook, we could use also Mendeley, and so on. Just as a contrast, I've developed the same map for Europe. Of course, the map now is rescaling Europe.

There are more publications than in the whole African continent, but I did the same. I spot those topics that have a substantial amount of publications from European researchers, and also those topics that have at least half of the publications with at least one tweet. And here we see a slightly different pattern.

So we identify topics like cancer, like psychological issues like anxiety, depression, bulimia, issues like autism or Asperger, so other topics as compared to the African profile. And somehow, to me, this relates to the presentation of Job this morning.

So we have topics like cancer, like psychological issues that are having a stronger attention on Twitter. So we can say that these tools, or these metric tools, allow us to identify these differences and these patterns. Another example is, for example, the communities of attention.

Again, I take the publications from Africa, and then I plot what we could call the Twitter coupling network. So essentially, in this network, what we have is that two Twitter accounts are together in the map, not because they follow each other, not because they know each other. We put them together in the map because they are tweeting the same publications. So they seem to have a similar interest

on the same group of papers. So then we can see a cluster of Twitter accounts that are related to HIV. The blue cluster is more related to environmental and ecological issues. And there are some other more multidisciplinary clusters. So the idea would be to select each of these clusters and then look at also the composition

of these communities of attention. Again, so this can inform who is tweeting, in this case, the African publications, how they are doing it, who could be relevant players, relevant users in this community, and somehow help us to build a story, to build a discussion about how, in this case, African publications are being discussed,

are being received on Twitter. But okay, then the second perspective is still how do we incorporate this more evaluative perspective in the use of our metrics. And here I would like to introduce a kind of dichotomy. So two different perspectives, two different foci,

even if we want two different worlds. So we have the scholarly perspective, the scholarly orientation, the scholarly focus. And here I would say we have something that we know pretty well. So in the top extreme, so the very scholarly focus, we have traditional databases, bibliographic databases, citation databases.

We also have peer review for evaluation. So systems that have their norms, that the scholarly community somehow are already familiar with the norms, with the rules, they know how they work. And then we have in this two perspective, the scholarly media focus and the scholarly focus, we have all the other new indicators

that we are essentially discussing in this conference. Somehow we can argue well indicators that come from Mendeley, that come from F1000Prime recommendations. Even I would argue Wikipedia citations have a conceptually speaking, have a stronger proximity to the more traditional, the more scholarly oriented indicators.

So Mendeley is mostly composed by academic related users, either researchers, professors, and also students. F1000Prime recommendations are done by scholars themselves. And Wikipedia citations are also typically provided by scholars. So we could argue that somehow these indicators are probably conceptually closer to what we know.

And I will argue later that these indicators perhaps could be included in the more traditional way of evaluating science. But then we have all the others. We have Twitter, we have Facebook, these are in the other extreme of the spectrum.

So basically they have their own rules that have nothing to do with the scholarly rules, with the scholarly norms. So the reasons why people tweet papers are multiple, are diverse, are actually difficult to grasp all of them. And somehow it's difficult to argue that we can incorporate these metrics in the more traditional,

in the more scholarly focus use of indicators. So in a way if we would summarize, we could divide the new metrics in two main big blocks. Of course we can subdivide them in many others, but at least two main big blocks. Those that have some conceptual similarity to citations or peer reviews,

or to the traditional ways of evaluating science. So here we would have readership in online reference managers like Mendeley or Sotero, or for example, post-publication peer review platforms. So here F1000 Prime, Puyier or Parmet Commons among others. And again, here I would argue that, well these are more suitable

for traditional research evaluations, so we could complement citations with Mendeley readers, with 1000, for example, peer review recommendations and so on. And then we have the other group, those that are dissimilar to citations, that conceptually are more different. So here I would say we have Twitter or Facebook mentions.

As such, and I'm not saying here always, but such as we are getting them from the most important and metric data providers, I would argue they are not meant for the more traditional way of evaluating science. Again, because the norms and the distance between these two worlds is too big. However, I wouldn't say that they are totally unusable

or they are totally irrelevant. From a different point of view, and this is the one I want to introduce, is that they provide a window on the social media dissemination or the social media reception of scientific publications. And here I would argue that if we consider that scholarly communication

is an important part of the creation of new knowledge, I would argue that then what happens about science in an important communication platform like social media, or social media platforms, it's also important. So basically what is happening on Twitter about science is non-trivial.

It may be that what is happening there is informing also the creation of new knowledge. So we can learn about concerns about groups of users, we can discover new topics that have a strong attention, and we may not be researching them or promoting them in perhaps the right way to solve questions, to solve concerns,

to solve multiple issues. So that means that if we look at that set of metrics that have this stronger social media component, we could discuss new evaluations if we also discuss new questions. So for example, the new indicators

could inform questions like, well, how is the output of my university being discussed on Twitter or in news media? Are the publications of my team or of my institution visible among the most relevant communities of attention? Are we really visible for those Twitter that are disseminating, discussing science?

Do these communities engage with the publications or they are just mere receptors of those papers? Is that reception, is that engagement with the publications positive, negative? Are we getting feedback? Are we doing anything with that feedback? And even units, researchers, groups, or other entities in my institution

active on social media, promoting, discussing, somehow even when clarifying the value, the relevance of our research. For example, what are the social media communication strategies at my university? Are they being effective? Are we targeting the right people, the right platforms?

Or for example, a much easier question is do open access publications help in the debates that are taken in social media? So just some final remarks to summarize a little bit my main points. I think we can argue that social media metrics are able to capture novel interactions between social media users and scientific objects

or scientific entities. And this is a whole new form of interactions that also opens a whole new range of research questions. We can argue that the social media reception of scholarly objects can be seen as a non-trivial aspect of scientific communication. So again, what happens there matters.

Almetrics and social media metrics can help us to monitor the coverage, density, and reception of scientific objects on social media. I would argue that they are a novel element in research evaluation, particularly when we regard scientific communication. So somehow we can inform issues related

with the public understanding of science. How is the science of my university being understood by different communities, if understood at all? So then there will be a kind of shift in the focus. So it wouldn't be that much on the scholarly impact in the traditional sense of the production of a research unit,

but rather on the social media reception or dissemination of its outputs and activities. And my final argument would be that somehow this idea of studying the relationships, the interactions between social media and science actually open a whole new research agenda that moves beyond the simple idea

of counting tweets or Facebook. So basically I would argue that we are walking towards an idea like the social media studies of science in the sense that science, and this is my last graph, science and social media have interactions. Somehow Almetrics so far has focused on this side of the relationship.

We have been looking at what happens on social media about science. I think we have also the opportunity to see what happens in science, what happens among researchers when they go to social media. For example, a very interesting question that I would like to answer at some point is, does the social media,

sorry, the symbolic capital of a researcher, say a highly-reputated researcher who goes to Twitter, does it have an impact? Does the researcher also gets all our followers and what he or she says on Twitter has also a strong visibility? And vice versa, that researcher that has a strong social media reputation, does it have an impact also on his scientific activities,

on his citation impact, but also on other types of scholarly impacts? And well, this is it, thank you. Thank you very much for having me. Okay, thanks everyone. So I'm Mitya Leacraft. I'm a marketing director

in the research marketing team at Spring in Nature. Whilst this is thinking, I will just carry on as if you can see more than you can see. So just to give you some context at Spring in Nature, we've been publishing open access for close to two decades now. So lots of experience in publishing open access,

but what I'm gonna talk about more today is how we're analyzing what it is that we do. So we have nearly 600 fully open access journals and nearly 2,000 hybrid journals. We also have a growing open access book publishing program and a growing set of journals and researcher services

that support open data. So we are really thinking about the open research story rather than open access by itself. We cover pretty much every academic discipline, both in broad multidisciplinary journals, such as scientific reports or nature communications, but we also publish a number of community titles. So 30% of everything that we're publishing at Spring in Nature is now open access.

So what I'm gonna talk about is why should we care or how should we do assessment of open access and open research? So for authors, we still hear an awful lot of the time that they don't really know what they're getting out of publishing open access. Or they have a vague notion that you gain some kind of extra visibility

and that equates to impact. So there's a lot that we can do in terms of thinking about how we can use metrics to start to tell a story about what they're achieving with their open access papers or books. And for funders as well, there are other issues that are a primary concern that they want to see from publishing open access.

So they want to see accelerated discovery and innovation. They'd like to see lower costs or increased ROI. And they'd like to see increasing data reuse. And I particularly like this quote from Charmaine from Cancer Research UK because I think this brings out all of the things that a funder would be expecting to see and why we should be providing more assessment

on what we're publishing. So what can our metrics do to contribute to this? Well, for broadness, we're not looking just at academic, reaching academic communities where we're thinking more broadly about what other communities are we reaching with our open research. And in terms of diversity, being able to look at multiple research objects

that has already been covered in this session. We're not just thinking about journal articles. We can think about open access books. We can think about open data. Multifaceted in the sense that we're looking at the same object from multiple lenses. So what kind of Twitter attention is it getting versus downloads or citations

and bringing all of those aspects together. And then lastly, speed. So we know open access is fast, but we're using our metric signals to actually be able to track attention much more quickly. And so bringing all of these aspects together enables us to start to tell a story around what publishing open access means for an author or for a funder.

So what we've been using at Springer Nature is a combination of downloads, whether that be article accesses or individual book chapters, citations, and then using online attention via our metrics as that proxy for attention more broadly and impact. So we have our metric badges integrated

across our journals platforms. And we also have book metrics which is a platform we've created jointly with our metric as a supplier for the data, but incorporating in-house data as well to help book authors really understand what they're seeing. So I'll show you some of that in context. So just to give you some idea

of how we're using this in context, one thing we've tried to do is start to show the overall benefits of publishing open access. So this is a report that we produced last November, which looks at the effects on open access books in terms of downloads, citations, and online attention.

And we looked at a comparative study here of closed and open access titles published by Springer and Palgrave Macmillan. And what we were able to show quite clearly is on average there would be seven times more downloads for those titles, 50% more citations. And in the first three years,

open access books would achieve 30 mentions, which is 10 times more than non-open access books. So all of this is, again, helping to tell that story to authors about what they're gaining from publishing an open access book. And we hear this when we then look to author testimonials. So this is Owen Davies, who is a professor at the University of Hertfordshire.

And he came and spoke at an event we held in April, which was really trying to tease out some of the comments that have been made today about really looking at the narrative behind those numbers. So for him, what he wanted to achieve was communication beyond academia and using his book metrics results,

he's able to see that in the first year of publication, he did achieve more than 5,000 downloads, but more importantly for him, he could see that he was achieving mentions and drilling down, he could see a lot of that conversation was happening on Twitter. So that backs up that he's achieved the aims of what he was setting out to do. Similarly, this is Dr. Rosalie Pellens,

who again spoke at our academic book week in April. And what she was trying to do, her work is in biodiversity, so there was a very limited period to influence policy. She wanted that to happen quickly. So for her, open access was about immediacy. And again, using the book metrics tool that was available to her, we can see that not only has she achieved more than 120,000 downloads

since the book published in 2016, but that bottom chart shows you the immediacy of how that was achieved through open access. So again, we're using these results to really help to back up and tell that impact story. Another report we've recently produced is looking at the same set of factors,

but for hybrid journals. So I doubt that there are many people in this room who are not aware that hybrids are a hot topic of conversation at the moment. There are a number of policy reviews underway. And the view for Springer Nature, and I'm sure we're not alone, is that there is merit to continuing to provide funding for publishing in hybrid journals,

both from a researcher perspective and also from a publisher and funder perspective. So we produced a study of over 70,000 articles, which tries to set in context again, what is the actual benefit and return of publishing in an open access hybrid journal? So this is just some of those results. These are the average figures.

We also took model data, which then would control for impact factor of those journals, the geography, the disciplines, and the institution that those articles were being published from. So there is more data in the actual report, but the averages again show that there are four times more downloads, 1.6 times more citations, and then looking again at the altmetric data,

2.5 overall in terms of altmetric attention scores, but that broke down to 1.9 times more news mentions and 1.2 times more policy mentions. So apologies to anybody who heard me bang on about this in July at the altmetrics conference, but we need to put all of this in context, and it's been so nicely set up

by other speakers this morning. We have to acknowledge that all metrics are flawed, and we should be considering them responsibly. There are lots of external forces driving change, so metric tide, the Lidar Manifesto, and the REF. We all should be considering those broader considerations of impact. And I absolutely love this graph, which is from a study

in Health Research Policy and Systems, a BMC title. It's a systematic review of health policy research, so the definitions of research impact. So this sets out what some of those look like, but they found 108 definitions in 83 publications. 108 definitions in 83 publications.

So that gives you an indication of the kind of breadth of landscape that we're talking about here. So for me, what that is saying is we need to dig deeper to really understand what successful open science look like. We've barely scratched the surface so far. I am personally excited about starting

to spend more time looking at qualitative indicators, so looking at who's reading that content, start to include other research objects, and look at real-world impact. So some of that might be public and patient engagement, some of that might be societal and economic impact. There's lots more work to do, and we're in the process of talking

to a number of partners about work that we can do in this area. So I look forward to share that with you in the coming year ahead. Thank you.