Bibliodiversity in transition-NISO Plus
Bibliodiversity in transition-NISO Plus
https://asa1cadmoremedia.blob.core.windows.net/asset-57cef9fd-e8c4-4dad-9edc-be0474c96cac/Bibliodiversity in transition-NISO Plus.mp4
GREG GRAZEVICH: Hello, and welcome to the NISO Plus 2022 session, Bibliodiversity in transition-- how open infrastructures support multilingualism, or not. I'm Greg Grazevich of the Modern Language Association of America, and I'm pleased to serve as moderator for this exciting and timely discussion. Let's begin with Jean-Francois.
JEAN-FRANCOIS NOMINE: Thank you. Thank you, Greg. A few words about me first-- I am an employee of the French National Center for Scientific Research. And I work in its research support unit called INIST-- so in plain words, Institute for Scientific and Technical Information. Basically, my background-- I'm a trained translator.
JEAN-FRANCOIS NOMINE: I'm a member of different working groups in my country dealing with open science. But to start this presentation, let me say that the point of view I will develop to enter our subject is that of a scientific and-- scientific and technical translator. So I will share my screen. So yes, translation-- translation, where to? is basically the backbone of this presentation.
JEAN-FRANCOIS NOMINE: To do their job-- oh, yes. So to do their job, it's not enough for translators known words in dictionaries write or correct machine-- write or correct machine translation, and done. We have to look for documents in the source and target languages in the field concerned, by the text to be translated, by framing our document research as efficiently as possible in connection with the workload inherent to the translation to be done.
JEAN-FRANCOIS NOMINE: Among the documents we search for, we are sometimes happy to find some for which there is a trustworthy, quality translation to adjust equivalent terms and phrases appropriately. Moreover, this makes us aware, in our own way, of the need of any person who is faced with a text in the language they do not command-- they do not comment and would like to read in their own language or in another one they grasp better.
JEAN-FRANCOIS NOMINE: But how is one to know if that translation exists? In the digital world of scientific literature, it happens that publication venues publish all or part of their articles in bi or tri or even multilingual versions. In this case, it is really relatively easy. A good search engine will do the trick. Somehow you will get to a text, and you will be able to see in the target publication environment that it has a translated version or an original in another language.
JEAN-FRANCOIS NOMINE: But this is still related to randomness and a relatively targeted need for information. In addition, translations are not only produced [INAUDIBLE].. They can be the result of people who identify documents well enough to decide to translate them or have them translated and to make the translations available in one form or another.
JEAN-FRANCOIS NOMINE: Not to mention large international organizations that have a policy of disseminating content in several languages at the same time-- UN, EU, WHO, World Bank, et cetera, more or less. Because producing translation, even with the help of technology, requires great effort. This is quite natural in international generic policy-inducing organization. But is that so when it comes to scholarly content?
JEAN-FRANCOIS NOMINE: Hardly partly, it seems. Now I would like to tell you a story. In November last year, I translated from English into French a guide on the use and usefulness of preprints. It was intended for researchers as well as to the general public-- the original words written by a group of Dutch researchers and librarians [INAUDIBLE] academic communities, and also to explain how to make good use of these preprints to a non-research reader.
JEAN-FRANCOIS NOMINE: The eminently practical content and the targets of this document justify the French translation. The work had already been done, most aptly. So why write something similar from scratch? Almost 90% of the content was well-suited to the French context, except for some minor adaptations that needed to be made to make the recommendations relevant to the French background. So at the end of the final phases of the translation, some minor adaptations were made to guide the French readership adequately.
JEAN-FRANCOIS NOMINE: The original document had its DOI. So the French document was deposited in HAL, the online archive of the French higher education and research. And its specific DOI was registered. Of course, at the time of registration, the corresponding metadata assigned to the translation status of this work was entered accordingly.
JEAN-FRANCOIS NOMINE: Well, every effort had been made to clarify that Les Préprints-- Guide Pratique, with its DOI, is the translation of A Practical Guide to Preprints, accelerating scholarly communication. And everything had been done to make sure that each of these texts could be found. But is that enough? It could well be. The search from the English title of the original on the [INAUDIBLE] search engine could well provide, in the list of the very first hits, a certain number of links referring to the archive which contains the French translation of this document or web pages mentioning this title, as well as a link to the [INAUDIBLE] about the publication of this translation and the indexing work done by the search engine.
JEAN-FRANCOIS NOMINE: Incidentally, digital machinery helps a great deal. Now, another story-- here is another document dealing with preprints. This one has been produced as part of Knowledge Exchange activities, a partnership between research organization in six different countries in Europe. Representing some interests-- interest, it's executive summary has been translated into French, published on the site specifically dedicated to open science of the French Ministry of Research and Higher Education.
JEAN-FRANCOIS NOMINE: And another version has been deposited again on the open archive of higher education and research in France, HAL. Another search with-- another search with the original-- with the original-- sorry-- English title leads to the page of the website dedicated to open science, commenting on this report, and linking to a French version of its executive summary. But it is not as easy as previously to find a version of this document registered in the archive.
JEAN-FRANCOIS NOMINE: The difference in the engine's behavior is probably the result of the fact that the English title string is placed within the engine's indexing reach in the first case and not in the second. Moreover, the web page found by the engine doesn't indicate in its information the content of the rest of the page and the exact rendition of the executive summary of the report it reports.
JEAN-FRANCOIS NOMINE: It only indicates the original English title. In the first example, as in the second one I just gave you, we are faced with a situation where the relationship between language versions of the same content could be facilitated. I'm unable to go into the specific reaches of the metadata practicable for the reporting of documents corresponding to this situation. However, I wonder about a possible solution to this problem.
JEAN-FRANCOIS NOMINE: There are services such as UNESCO Index Translationum, but this one seems to focus more on literary or cultural works. In the past, there was the world discontinued for many years now. The [INAUDIBLE] UNESCO service is currently under maintenance. But we can be sure that it will resume its activity in the short term. However, as a layman, I wonder about the technical means that could be used to provide a specific service that would automatically index technical and scientific documents that have different versions in other languages, whether they are full, partial, or adapted translations with the means to indicate this variance.
JEAN-FRANCOIS NOMINE: The two examples I have just given you are undoubtedly evidence of clumsiness and ignorance on my part from an information-illiterate speaker. Granted, I learned in the meantime there are differences in-- the differences in scope and use between Crossref and DataCite concerning the assignment, that frontiers are not very clear in the minds of users, and that, in particular, their schemas address translations a little bit differently.
JEAN-FRANCOIS NOMINE: But reading about these identification possibilities and the technical details caused me to wonder if something could be done based on these very helpful services-- and something smooth and easy for a layman like me. And if I were not the only one, would this be a reasonable assumption to start with in view of developing servicing bringing scientific and technical translation to the fore in our digital work environment.
JEAN-FRANCOIS NOMINE: We might live in a world of an increasing number of, so to speak, lingua francans-- that is, people who have a good command of two languages and less so, but still so, in some others. Wouldn't it be nice to help them better grasp documents when they detect them without being able to further pore over their content beyond superficial understanding of what they are about? In addition, as machine translation is gaining traction in scientific writing and communication as an assistive technology, post-edited, reviewed, and trusted machine translations of text may increase in number in the future are, in fact, very simple.
JEAN-FRANCOIS NOMINE: Is there a way to make objects that are complete, partial, or adapted translation go to the surface in our digital universe? Can this be done by associating elements of trust or indices of reliability to the information produced? Can technology make this easier? Or is a service conceivable, at least in the scientific and scholarly field? What is conceivable?
JEAN-FRANCOIS NOMINE: And finally, another questioning, less technological, but economic and human-- what sustainable systems could be put in place? How could one provide for their sustainability? This would perhaps contribute to multilingualism in platforms if it could be done. Thank you very much for your attention. And now I think I'm to hand over to Suzanne.
SUZANNE DUMOUCHEL: Thank you, Jean-Francois. Thank you very much. So I'm Suzanne Dumouchel, and I work at CNRS, where I am the Scientific Coordinator of the Triple project that I will present you briefly. I'm also one of the two coordinators of OPERAS Research Infrastructure and part of the board of directors of the EOSC Association, the European Open Science Cloud Association.
SUZANNE DUMOUCHEL: So let's start about the Triple project, where I'd like to discuss how we address-- we try to address the multilingualism topic in the discovery platform named GoTriple. So GoTriple actually is a discovery service OPERAS Research Infrastructure. OPERAS is a European infrastructure with a legal entity. It's dedicated to open scholarly communication in the field of social science and humanities.
SUZANNE DUMOUCHEL: And it provides several services for its community. Their infrastructure is organized in Special Interest Groups, which constitute the Assembly of the Commons at the General Assembly. And it promotes largely multilingualism, especially through its services and its national nodes that constitute the OPERAS Executive Assembly, and also through a Special Interest Group which is dedicated to multilingualism.
SUZANNE DUMOUCHEL: So now, what about the Triple project. So Triple is a funded project funded by the European Commission. It stands for Transforming Research Through Innovative Practices for Linked Interdisciplinary Exploration. It has been launched in October 2019, with 21 partners in Europe, coordinated by CNRS in France, with a duration of 42 months and a budget of 5.6 million euros.
SUZANNE DUMOUCHEL: What is Triple? So GoTriple is a multilingual service. Nine languages-- French, Spanish, English, German, Greek, Portuguese, Croatian, Italian, Polish-- are part of the platform. It has been decided to focus on languages less present in the International exchange of knowledge and to confront the difficulties in non-English-speaking circuits.
SUZANNE DUMOUCHEL: And this is why we used openly available resources to facilitate the uptake and the reuse of the work we are doing. GoTriple focuses on subject-specific and multilingual diversity on the one hand. And on the other hand, it enables researchers to identify not only data but also other scientists and project in the European area. This is why we have the Triple for. You can see here data publication, profile, and project.
SUZANNE DUMOUCHEL: So the topic is to facilitate the linking and the discovery of this kind of resource in Europe and across the subject and language boundaries. So how do we address the multilingualism in GoTriple? Actually, there are three ways that we use to address the multilingualism issue in GoTriple. The main ways-- there are some other, but I'd like to focus on this one and especially on the bottom-- the one at the bottom, the vocabularies alignment, which I will present a bit in detail, much more in detail, actually, rather than the two others.
SUZANNE DUMOUCHEL: So what is interesting here is that the vocabularies alignment, actually, is the most-- more technical aspect of the work for addressing multilingualism in GoTriple. But it's the less visible for the users. Even if it's the less visible, we spend half of the project on it. And I will explain why, what kind of process and methodology we employ to work on the alignment of the vocabularies.
SUZANNE DUMOUCHEL: So this aspect is really linked to the search and the core platform of-- its behind the interface for the users. And then you have the two other ones-- the one with-- the second one, which is showing multilingual metadata, which is actually as the way in progress. We are currently working on it, so I will be able to show you some prefiguration of what it will be in the final version, and the interface available in several languages, which is still work to be done.
SUZANNE DUMOUCHEL: So these two last ways are much-- are less technical rather than the other-- the vocabularies alignment, but much more visible for the users. So let's start with this first one, the vocabularies in SSH and their alignment. So what kind of process did we establish to determine results to classify? But I'm not really happy with the word "classification" [INAUDIBLE] "classification" because it's related to something very specific in our project.
SUZANNE DUMOUCHEL: So how do we work further to determine a resource in the GoTriple platform? First, we established a classification by discipline, where this resource is-- to which discipline this resource is linked, is part of. And for that, we use the MORESS categories. So there are 27 SSH discipline part defined in the MORESS project. And then we classified the results by the language used in this resource.
SUZANNE DUMOUCHEL: The second step is the annotation of the resource by tagging item or metadata in one of the nine languages. This annotation, it implies or allows to have more searchable research for audiences that speak any of the other languages in the platform. And then the third step is the enrichment with links that help to-- in the disambiguation of the term and increases its semantic interoperability.
SUZANNE DUMOUCHEL: So here, we will focus on the second step only, the annotation of the resource, which is the one linked to multilingualism. But this is a way for you, in this slide, to understand how multilingualism must be addressed in other complex processes in order to be then reusable and-- yes, reusable and useful for the researcher in the different platform.
SUZANNE DUMOUCHEL: But then what is a controlled vocabulary? Why do we need to have this kind of thing to work on this kind of tool in the GoTriple platform? So controlled vocabulary is an organized arrangement of words and phrases used to index content and/or retrieve content through browsing or searching. The purpose of controlled vocabulary is to organize information and to provide terminology to catalog and retrieve the information.
SUZANNE DUMOUCHEL: And we need, in social science and humanities, a reliable vocabulary. Why? First, because the community is very fragmented. The social science and humanities researchers are often-- they publish-- they can publish in their native language. They are often dealing with very cultural or specific geographical area.
SUZANNE DUMOUCHEL: And it's a very fragmented community, because of the 27 disciplines that are part of the domain SSH. So this is why we need to work to propose reliable social science and humanities vocabulary which [INAUDIBLE] which allow to have better access to the resource in-- to find better resource. This is why having reliable vocabulary is essential. It's necessary at the indexing phase, because without them, the catalogers will not consistently use the same term or refer to the same concept.
SUZANNE DUMOUCHEL: But it's also necessary in the retrieval process, because the end users may use different synonyms or more generic terms to refer to a given concept. And this is exactly why GoTriple is not only a tool or a platform useful for the social science and humanities researcher, but also for other researchers which are not aware of the right term used in certain disciplines from the SSH but can find it thanks to the SSH vocabulary. And this is also why it's also a tool that can be used for citizens and societies at large.
SUZANNE DUMOUCHEL: So to ensure an effective annotation mechanism for publication of all the nine languages in GoTriple, the vocabulary must contain a sufficient number of concepts. And at the same time, these concepts must have labels in as many of these languages as possible. So how did we do-- how did we do that? Five steps to build the GoTriple vocabulary-- the first one, existing classification, and so we gather and compare existing classification and controlled vocabulary.
SUZANNE DUMOUCHEL: The second one-- from this first analysis, we create new vocabularies which were required for the description in new languages. Then we select-- we selected one existing vocabulary which was-- which were used as a basic resource in English for the construction of the Triple vocabulary. And for that one, we use the Library of Congress Subject Headings System, which is probably the most widely adopted subject indexing system in the world and which has already been translated into many languages.
SUZANNE DUMOUCHEL: Then we move to the first step, the automatic translation of the SSH concept in each of the nine languaging-- languages, sorry. And the last concept, the human validation of the translation by experts of each language to confirm the translation, the automatic one, or to suggest a better one if needed. Then let's focus on the selection of concept. So we started from the Frascati taxonomy.
SUZANNE DUMOUCHEL: We use that taxonomy under social science and humanities to identify 14 basic concepts. Then we enlarged them to arrive to 37 broad concepts from the LCSH vocabulary. And for each of these 37 broad concepts, we used the link data API of the Library of Congress and retrieved their SKOS representation. As a first step, we followed the skos.narrower property. So that means, actually, the concept that we are very close-- already identified as very close of the previous broad concept.
SUZANNE DUMOUCHEL: And we extracted their children in SKOS. That means we ended up with 2,513 concepts. And then we increased the multilingualism of the vocabularies by three ways. First one, following the links to LCSH; the second one, following the links to wikidata and by extracting [? labels ?] in various languages; and the last one, by ingesting existing mapping from national vocabularies-- for instance the French and Italian National libraries.
SUZANNE DUMOUCHEL: So that means the multilingualism of the = has been increased further, then, by using automatic translation services to produce missing [? labels ?] which then were validated and/or curated by partners. So here you can see an example of spreadsheets. Actually, we had eight spreadsheet, which were where all the automatically generated translations were validated and curated by partners.
SUZANNE DUMOUCHEL: So we had two-step validation in order to secure the automatic translation. And now I can show you what is the current coverage per language. So you can see the evolution from October 2020 to June 2021. And you can see that now we have quite good results in the coverage of all the nine languages we have in the GoTriple platform. So what is interesting here is that it's a vocabulary.
SUZANNE DUMOUCHEL: So that means it must be adapted depending on time, context, actuality. So we have already a good starting point. But it will need to be updated depending on the evolution of the concept and the research in social science and humanities. So let's move now to the second way of adding multilingualism in GoTriple. So as I said, this step is actually one in progress.
SUZANNE DUMOUCHEL: So what I'm showing you now is only a screenshot of the design, the beta-- yeah, beta version of the platform but they are not already available. Nominally, it's planned in March 2023, so quite soon actually. So we had several discussions in the Technical Board to see how to proceed with metadata translation-- how to highlight the-- how to show, not highlight, but how to show different translation of title, keywords, or even the abstract, depending on the language spoken by the user.
SUZANNE DUMOUCHEL: So we decided to have this kind of scrolling menu, as you can see here on the left. And now you can see that we have a title in English, with the keywords in English and the abstract in English. And the result now is here, where everything has been translated in Polish, because it's probably most easier for the researcher. So this is where we are. And we are still working, as I said, on this aspect.
SUZANNE DUMOUCHEL: And lastly, the whole gotriple.eu interface-- so actually, it's not yet there. It's planned. To be very honest, it was not planned at the beginning of the project. We decided not to translate the GoTriple interface in other languages rather than English. But we had a lot of feedback from beta testers from the survey and the co-design workshop we made, where people told us, well, it's not a multilingualism interface, because everything is in English on the page.
SUZANNE DUMOUCHEL: OK, so we tried to explain to them that it's not exactly how we think what is multilingualism. But we understand that it was not really what they expected. So because we are completely on time in the project, we decided to start a co-design activity, where the translation of the interface in several languages will be made by participants of a future hackathon planned in 2022 and with the support of the OPERAS national nodes.
SUZANNE DUMOUCHEL: And actually, the translation of the interface had already been sought in the development of the platform. So it will be very easy to implement the result of the different workshops. And to conclude, I just would like to say that, in the Triple project, we tried to reconsider standardization and diversity with a kind of methodology, but also to highlight the diversity and multiculturalism of-- that we have in social science and humanities.
SUZANNE DUMOUCHEL: And for doing so, we nurture wikidata with concepts in different languages, we highlight regional diversity of such resources and keywords, and of course, the methods and processes that I just presented will be used for adding new languages. And they can be reused for other initiatives in order to promote multilingualism. And with that, I'd like to thank you very much for your listening.
SUZANNE DUMOUCHEL: I'd like also to thank my colleagues, Iraklis Katsaloulis from EKT and Cezary Risinski from IBL Pan, because I used some content of their slide presented at the first Triple International Conference. Thank you very much, and now I'd like to hand over to Abel for the last presentation.
ABEL PACKER: Thank you very much, Suzanne. Congratulations for the presentation-- very interesting. We are very interested in GoTriple [INAUDIBLE].. And thank you so, Jean-Francois-- very pleased to participate in the same meeting here. Thank you, Greg, for introducing us, and fantastic [INAUDIBLE] organizing this meeting. OK, I will share with you our biggest challenge we face in SciELO, particularly SciELO Brazil, regarding language is a major challenge.
ABEL PACKER: So we don't have-- not yet solved [LAUGHS] the problem. But we are now more aware of the [INAUDIBLE] of the problem. So let me do my presentation. So SciELO Network is, naturally, a multilingual platform and multilingual [INAUDIBLE] Portuguese, Spanish, and English.
ABEL PACKER: I tried several times to introduce French, but we were not capable to go ahead. So it is a network [INAUDIBLE] 17 countries with [INAUDIBLE] We have a centralized collection of public health [INAUDIBLE] And moving to open science, we are running from [INAUDIBLE] SciELO [INAUDIBLE] and SciELO data repository.
ABEL PACKER: So we are learning how to manage these open science practices. And we have also a collection of SciELO Books practically 100% in Portuguese now and, progressively, also in Spanish. If you will look at the network, I took it 2020 15 countries, 300 journals, 54,000 articles.
ABEL PACKER: These are the distributions-- [INAUDIBLE] So we have it in English. Life science and physical science and social science. But it is more in Spanish and Portuguese. This percentage go over the entire network. And also, Spanish is more frequent than Portuguese, but the behavior is similar. What is very interesting is the multilingual, which is the publication of articles simultaneously in two or even three languages.
ABEL PACKER: This is same data, but looking out for major [INAUDIBLE] We can see that in life science has this behavior. Same in physical sciences. OK, so more important today, 2020, I would like to show you the evolution of a language. What do we see in Latin America is that we have a Spanish practically constant.
ABEL PACKER: In Latin America, except Brazil, they continue to publish the same proportion of Spanish. And we have, in all cases, an increase in English, a decrease in Portuguese, which is mainly promoted by Brazil in all areas. And we have a slow increase in multilingual publishing, basically Portuguese and English.
ABEL PACKER: When you go to Brazil, we can see the difference, because Portuguese goes from 70% to [? 46%. ?] And English goes from-- is absolutely inverse here. And we have almost 23% which is published simultaneously in Portuguese and English, which is a big effort, technically.
ABEL PACKER: So to do the translation and [INAUDIBLE] So let's go to see one example. This is the home page of Brazil. I will go to take in the alphabetical list this journal, [NON-ENGLISH]. This is the collection where you take the 2020 supplement one is an article [INAUDIBLE].
ABEL PACKER: Social distancing measures to control the COVID-19 pandemic-- potential impacts and challenges in Brazil is an article, research article. This journal publishes simultaneously 100% of the articles in Portuguese and in English. So we have the abstract. We have the text, HTML, which are-- which is produced from the XML, and also the PDF in English and Portuguese.
ABEL PACKER: So this is the operation that we do in the SciELO platform, OK? So we have the same in Portuguese, so the same-- you can look in Portuguese or in English. When you have a Spanish [INAUDIBLE] you can see also the same distribution in Spanish. OK, let's look at the article here. You have, of course, the abstract here. And you can see, over here you have the abstract in English and Portuguese, text in English and Portuguese, PDF in English.
ABEL PACKER: Here is the version in Portuguese. And so here is the text. I won't show in English, but you have it in Portuguese, also. Then you have the PDF import in English, OK? You can see that we have only one abstract. Then we have [INAUDIBLE]. So we have a title in Portuguese and always the translation in English, abstract and the keywords, OK?
ABEL PACKER: So if it is in English, it goes metadata [INAUDIBLE] English. If it is in Portuguese or Spanish, it's mandatory translation, title, abstract, and the keywords to what we do. But [INAUDIBLE] the index's system do not process that. So they are penalized [INAUDIBLE] So let's go. Here is a search of this article in Google.
ABEL PACKER: So we have four major versions here, with 189 citations-- very popular article. Then you have a order in English, in Google, with 183 citations. We never know if the citations are matching or not. Nobody tells. Here is Google's Scholar, which is better.
ABEL PACKER: let's say. organized than the Google itself. So we have three. But we have here 10 versions of this article. But Google is the best system in making the aggregation of the version. You have one with 89 citations. [INAUDIBLE] the same copied by Google, from Google Scholar.
ABEL PACKER: When I do-- I did a search in English here. When I do it in Portuguese, I got 408 [INAUDIBLE] The impact in Portuguese is different of the impact in English. [LAUGHS] I am asking [INAUDIBLE] What did this mean, the impact of the research? Who is [INAUDIBLE] this. If I go to the National Library of Medicine, which has a very sophisticated system, it does recognize that this article is in English and Portuguese.
ABEL PACKER: But it does not publish the title. It does provide a link to SciELO, because we provided the metadata here. The abstract-- they do [INAUDIBLE] there is abstract in Portuguese. And if you click here, you can actually read the abstract in Portuguese.
ABEL PACKER: But when you do not find it, even when they have all the data there, this is [INAUDIBLE] which is a major problem. We have [INAUDIBLE] today the data. And if I look [INAUDIBLE] I have this version in Portuguese. We don't have the version in English. [INAUDIBLE] you can see that they have the data structured.
ABEL PACKER: But you don't have the abstract in English, and you do not have the title in English [INAUDIBLE] So we will need to do a lot of [INAUDIBLE] changes to provide the [INAUDIBLE] with all the data, which means basically provide [INAUDIBLE] So if I go to Web of Science here, I can search it in English but not in Portuguese, even when they receive all the data.
ABEL PACKER: The only system that actually provide the access in Portuguese and English does not matter which language you search is [INAUDIBLE] I don't know how they do [INAUDIBLE] But today, they can [INAUDIBLE] So I search here Portuguese. I get the article. They display it in English. And when I search it in English, I get it [INAUDIBLE] So just to show how important is that [INAUDIBLE] I use it here because, in the case of SciELO, because [INAUDIBLE] the impact by citations is critical in our project, we can see that, even when we do everything to improve visibility of [INAUDIBLE] et cetera, the impact-- let's say the domestic impact-- is, in some way, exhausted already, OK?
ABEL PACKER: And as we increase the publication in English, we decrease the impact, the domestic impact. But we gain a lot, of course, in international impact. If we go to look for download, which we do very sophisticated with the COUNTER methodology, and we-- I took here documents that-- from 2016 and 2017 and downloads done by humans, following counter, uniquely requisitioned-- very sophisticated.
ABEL PACKER: And then we can see that the humans [INAUDIBLE] health science, which is expected, and social-- applied social sciences. And the biological sciences, which has the biggest impact in citations, they do rank last here. So we are improving this study here. So the question is, which impact do you want so you have your options to get it?
ABEL PACKER: So to [INAUDIBLE] my presentation, SciELO needs to improve its operation of the new English documents to increase the visibility. We will need to do that to improve our way to manage and to provide the data to the index. Global indexes, particularly Crossref, lack proper operational multilingual documents, what penalized very much in our case.
ABEL PACKER: And multilingualism is inherent to research communication, research scientific knowledge management. education, and many other aspects. But if the goal of research in journals in non-English contexts is impacted by citation, or preferably national [INAUDIBLE] language-- but if the impact is [INAUDIBLE],, of course, you can decide in the balance what you publish in non-English and English language.
ABEL PACKER: As I said in the beginning, we are more and more aware of the complexity of the problem, how SciELO is penalized. When you look to the citations, you receive Portuguese citations, you receive in English. We need to go, of course, one by one in each system to check if they are counted correctly. So it's a major challenge we have.
ABEL PACKER: So thank you very much for the opportunity.
GREG GRAZEVICH: Thank you, all. NISO Plus attendees, please click through and join us in a live conversation with the panelists with your questions and comments. [MUSIC PLAYING]