right now, i'm not 100% that what the dashboard currently does actually needs a shiny runtime.

I'm going to have to look into the current features of crosstalk etc. to see whether this can be done.

If we can get ~80% of what we want without a shiny runtime, that might be worth considering, because it would dramatically simplify deployment, cut costs and increase speed.

Not sure this is relevant, but as an open source fan but bibliometric noob, I wondered how the hybrid journals compared to full open access (and foss) journals.
Might be nice to see the trends over time in both and compare the magnitudes.

Maybe this isn't relevant or interesting to the expert audience.

general solution in subugoe/metar#35

as per https://github.com/ropensci/software-review as far as applicable.

there might be some duplication here

I would suggest showing the journal name in a separate widget, either via crosstalk or other linked brushing.
And I think we can disable the color.

Hi ,

i am trying to create the multiple dashboards but not able to understand how i need to populate my data which is in rscript , also how to include selection buttons as per my tables,
i have seen many examples but getting confusion.
please help me to create.

I am not sure what the status of this dashboard is, but it seems like they tried something similar.

Are there other projects we should consider to avoid code duplication?

I noticed this upon importing the citation into Zotero and can't get at it with a PR ;-)

also necessary for #51

aus dem Antrag:

die so gewonnenen Affiliationen werden danach mit der Liste der beitragenden Einrichtungen der Open APC Initiative abgeglichen. Die Güte des Retrievals wird mittels einer Zufallsstichprobe qualitativ überprüft. Die Arbeiten zur Datengewinnung und -kuration, insbesondere die Vereinheitlichung der Einrichtungsangaben, werden durch eine wissenschaftliche Hilfs- kraft unterstützt. Sollte das automatisierte Verfahren aufgrund heterogener Einrichtungsangaben sich in der Kalibrierungsphase als zeitlich zu aufwendig herausstellen, wird im Projektverlauf anstelle der projek- tierten Vollerhebung eine Zufallsstichprobe gebildet.

aus dem Projektantrag:

Neben dem Fokus darauf, wie Verlage Handlungsempfehlungen zur Kostentransparenz umsetzen, wird auch der Abdeckungsgrad der Open APC Initiative analysiert. Zum einen ist es Aufgabe zu überprü- fen, wie umfassend Open APC das hybride Open-Access-Aufkommen ihrer teilnehmenden Einrichtun- gen abdeckt. Dies kann als Indikator dafür dienen, ob verteilte Zahlungsströme an diesen Einrichtungen existieren. Zum anderen sollen Einrichtungen identifiziert werden, die derzeit noch nicht an der Initiative teilnehmen.
Die Datenbasis bilden erneut diejenigen Open-Access-Artikel in subskriptionsbasierten Journalen, die zu Beginn des Arbeitsschritts mittels der Crossref-Lizenzinformationen gewonnen wurden. Es wird zunächst die Differenzmenge gebildet, um Artikel ohne Nachweis im Open-APC-Datensatz zu erhalten. Sie um- fasst laut gegenwärtigen Erkenntnissen aus unseren Vorarbeiten rund 82.000 Artikel. Anschließend wird ein automatisiertes Verfahren entwickelt, mit dem sich je Artikel die Kontaktdaten (Emailadresse) und die Einrichtungszugehörigkeit der Corresponding Authors über Crossref, oder, wenn nicht vorhanden, über den öffentlich verfügbaren Volltext extrahieren lassen. Die so gewonnenen Affiliationen werden danach mit der Liste der beitragenden Einrichtungen der Open APC Initiative abgeglichen. Die Güte des Retrievals wird mittels einer Zufallsstichprobe qualitativ überprüft. Die Arbeiten zur Datengewinnung und -kuration, insbesondere die Vereinheitlichung der Einrichtungsangaben, werden durch eine wissenschaftliche Hilfs- kraft unterstützt. Sollte das automatisierte Verfahren aufgrund heterogener Einrichtungsangaben sich in der Kalibrierungsphase als zeitlich zu aufwendig herausstellen, wird im Projektverlauf anstelle der projek- tierten Vollerhebung eine Zufallsstichprobe gebildet.

I am wondering whether the canonical object hybrid_df shouldn't be a list of several dfs, rather than one with duplication.

just a small thing *.rproj currently has a different name than the repo, just trips me up when opening

this necessary for #51

Minor issue: Maybe you should make more clear that the lower right plot only relates to the yellow OpenAPC share of the current selection, not the whole set of articles - one might draw some very wrong conclusions here.

Otherwise, a great project!

... maybe.
would extend #28 and depends on #25

E.g.
https://api.crossref.org/works?facet=published:*,license:*,publisher-name:*&filter=issn:0586-7614,from-pub-date:2018,type:journal-article

Publisher page
https://academic.oup.com/schizophreniabulletin/issue/44/suppl_1

aside from tests of the shiny ui in #8 it might be worthwhile to add sanity checks to the scripts.

aus dem Antrag:

In Form eines Dienstleistungsauftrag sollen die Ergebnisse der Analysen mit der Scopus-Datenbank des Kompetenzzentrum Bibliometrie abgeglichen werden. Im Mittelpunkt stehen Fragen nach der Indexierung der Journale in Scopus, des Publikations- typs, der Fachzuordnung, des Publikationsjahrs und der Einrichtungszugehörigkeit der Autorinnen und Autoren. Scopus wurde gewählt, weil die Datenbank im Vergleich zum Web of Science über eine breitere Abdeckung an Journalen verfügt.

as suggested by @dirkpieper

https://subugoe.github.io/hybrid_oa_dashboard/about.html states:

For being best represented in this dashboard, publishers will have to make sure to include license URL element license_ref and a start_date equal to the date of publication in the licensing metadata, which helps to identify open access journal content as well as to differentiate between immediate and delayed open access.

Crossref defined "start_date" as "optional", meaning that the license is applied to the published resource immediately if no start_date is provided, see https://github.com/CrossRef/rest-api-doc/blob/master/funder_kpi_metadata_best_practice.md#license-information. Is there really a need to expand this recommendation?

general solution in https://github.com/subugoe/metaR/issues/29

might be via something fancy, or just regex.
should be a separate ETL project

just a wild guess, not even sure this is possible with the data, but if it is possible, might be interesting to look at it.

this is just a really small thing, but hybrid_oa_dashboard is getting a bit long to type.

Would it be alright to rename the (nascent) package and repo hoad @njahn82?

All existing URLs etc. would remain active or be forwarded.

available::available('hoad') suggests it should be alright:

Name valid: ✔
Available on CRAN: ✔ 
Available on Bioconductor: ✔
Available on GitHub:  ✔ 
Abbreviations: http://www.abbreviations.com/hoad
Wikipedia: https://en.wikipedia.org/wiki/hoad
Wiktionary: https://en.wiktionary.org/wiki/hoad
Urban Dictionary:
  [heart] of a [dragon]
  http://hoad.urbanup.com/11027756
Sentiment:???

as per the plan in #46

@njahn82 if you could point me to the source for the grant application, I'd like to also add that as a vignette to the project website.

with milestones etc.

as I mentioned, I'm going to need admin org access.

if not #24
modules should make the code more manageable.

aus dem Projektantrag:

Auf Grundlage des Datensatzes, werden zunächst die Crossref-Metadaten dahingehend untersucht, ob und inwieweit sie die in AP 1.1 ermittelten Berichtsdimensionen abdecken. Verlagsseitig ermöglicht das Crossref-Metadatenprofil Verlagen bereits jetzt, wichtige Informationen zur Gewährleistung der Kosten- transparenz zu dokumentieren. Zu ihnen gehören Lizenz- und Förderinformationen sowie Angaben zu Autorinnen und Autoren einschließlich der ORCID und ihrer Einrichtungszugehörigkeiten.
Es ist allerdings unklar, welcher Anteil an Journalen, die nachweislich Open-Access-Artikel nach dem hybriden Modell publizierten, die Möglichkeiten des Crossref-Metadatenprofils vollumfänglich und quali- tätsbewusst nutzen. Um diese Frage zu beantworten, werden im Arbeitsschritt die Abdeckungsraten für die relevanten Metadatenfelder erhoben. Für die Bewertung der Güte der Metadaten werden der Re- call (Wahrscheinlichkeit, mit der die entsprechenden Verlagsangaben über Crossref verfügbar sind) und die Precision (Wahrscheinlichkeit, mit der diese Informationen korrekt über Crossref abgebildet werden) ermittelt.

this actually has a pretty high priority, because we don't want to loose any historical data.

https://rstudio.github.io/shinytest/articles/ci.html

docs/ can be convenient, but has a some disadvantages:

• bloats the repo
• does not play nice with ci/cd

This is already largely the case, ETL is in R/, plotting in the dashboard.

I think the separation of concerns might go further here, by making functions for both.
ETL functions would then return a hybrid_df object, which plotting functions can, well, plot.

if possible, I'd like to put in a simple redirect from the shinyapps.io url https://subugoe.shinyapps.io/hybridoa/ to https://subugoe.github.io/hybrid_oa_dashboard/.
On https://subugoe.github.io/hybrid_oa_dashboard/ the existing shinyapps.io instance would then be included as an iframe.

Advantages:

• we can present everything from one URL which we control
• if we ever leave shinyapps.io, no one will notice or need to change anything
• the old shinyapps.io, under which this was first publicised, remains active
• we can implement full-on branch deploys as per https://github.com/subugoe/hybrid_oa_dashboard/issues/56 easily

There's an issue with the journal dropdown selection: The list gets cut off at 1000 entries (Happens when publisher is set to "all" or "Springer Nature").
To save you the first 100 meters of the rabbit hole: The reason is that selectInput utilizes a JS library called selectize.js in its default work mode and that library limits the select box options to 1000 per default ("maxOptions").

readme.md etc.

This might be too early now, though deciding this sometime soon might help with some other decisions (especially reproducibility and testing).

If we end up using a shiny app (and not #24), I see three possibilites:

1. RStudio Connect, hosted on-prem at GWDG.
   License is expensive, but has lots of good stuff, especially if we want to roll this out to external customers.
2. shinyapps.io (status quo, easy to do, though somewhat limited reproducibility and no support for plumber, if needed).
3. roll-your-own based on shiny server open source, maybe via shinyproxy

It seems that 31 journals are affected:

this involves some combination of

• DESCRIPTION
• {renv} #99
• {sysreqdeb} #95

I am wondering whether we should, as far as possible, replace the database dumps (i.e. *.json or *.csv) with (properly cached) queries.
I think this might aid reproducibility and make the whole thing more scalable, and automatically updated etc.

I don't know nearly enough about the license, auth and rate limit of upstream data sources, but if necessary, with a proper cache or even our own data base, this should be possible.

We could then expose a hybrid_df() function, which returns a tibble (or list of tibbles) from the query.
That would also be a nice place to document all of the details of the fields etc.

We might then, in turn, expose this ETLd query as a plumber API, which we could a) use ourselves in the dashboard and b) other people could use, even if they don't necessarily like or use R.

This would be a nice design where we fully separate the concerns between ETL and analysis/plotting, etc. both in terms of functions and infrastructure. (see #43).

The ETL functions would then drive the plumber API.

The plumbing and analysis functions would drive the dashboard.

Unpaywall Data is a great and popular source to identify hybrid open access journal articles. Although Unpaywall also makes heavy use of Crossref, methods and results differ. Unpaywall checks not only Crossref, but also publisher websites directly to find OA articles with open licenses. Contrary to our approach, Unpaywall Data does not indicate when an article was made OA with an open content license. Consequently, we cannot distinguish between immediate and delayed OA provided by toll-access journals using Unpaywall Data alone. However, because not all publishers share license metadata with Crossref and because some journals make articles available after a certain period like the Journal of the American College of Cardiology, we thought that integrating Unpaywall Data is useful to explore the extent of OA articles that is currently provided by toll-access journals in our sample.

We used the most recent Unpaywall data dump from September 2018. R/unpaywall_integration.R shows how we queried the dump, which we stored on Google Big Query before, for OA journal articles where Unpaywall found license information. The script also shows how the indicator jn_y_unpaywall_others was calculated, which is now present in our dataset data/hybrid_publications.csv. Results are presented on the page "Overview", tab "Other types of OA license information detected by Unpaywall".

First draft live with commit subugoe/hybrid_oa_dashboard@b53d851 . We need to add this to the long-form and data documentation.

I think PBJ switched from closed to open in 2017:

"Welcome to the first issue of the fifteenth volume of Plant Biotechnology Journal. I would like to start this editorial by announcing the successful transition of PBJ from a subscription‐based journal to an open access journal supported exclusively by authors. This resulted in enhanced free global access to all readers." -- https://doi.org/10.1111/pbi.12687

I'm thus surprised to see that it's in the hybridOA table with 187 articles out of 202 in 2017 (92.57%) for the year. Could this be an error such as including 2016 publications, which might account for the 15 not OA articles? It just seems odd to me.

Just for clarification: There are cases where the article count in OpenAPC is larger than the count reported by the monitor (Wiley-Blackwell 2014/15 or SpringerNature 2015). This means that your API calls could not identfiy those missing articles as hybrid OA in crossref because they are not correctly tagged with a CC license in license_ref, correct?

This would simplify dependency management, documentation, testing, etc. and might just generally be a good scaffolding.

Observations: