Wikisource Satisfaction Survey 2021

From Meta, a Wikimedia project coordination wiki
Jump to navigation Jump to search
Overall satisfaction across the 5 projects excluding the users who mentioned that they haven't used the tools.

The Wikimedia Foundation ran Wikisource Satisfaction Survey 2021 in July 2021 to understand the satisfaction of the Wikisource community for a set of projects associated with WMF. This included:

  • The Ebook Export Improvements and OCR Improvements projects undertaken by the Community Tech team as part of the Community Wishlist Survey 2020
  • Two projects (improving IA-Upload tool & Wikidata-Wikisource integration) undertaken by grantees funded by the WMF
  • Wikisource Pagelist Widget developed by a Google Summer of Code 2020 student

The survey was conducted in 6 languages and got 69 responses from Wikisource contributors from over 12 different language communities.

High-level findings[edit]

  • Amongst the respondents who had used the tools, there is an overall satisfaction of 75%, 15% of the users were neutral about the tools and 10% of the users said that they were not satisfied.
  • Amongst all the respondents, an average of 46% reported not having used the tools previously.

Ebook Export[edit]

For more details read: Community Tech/Ebook Export Improvement

Response No. of respondents Chart
Satisfied 34
Ebook Export - Wikisource Satisfaction Survey 2020.svg
Neutral 7
Not satisfied 3
Haven't used 25

Satisfied respondents[edit]

[en]

I have used the Ebook download tool on numerous occasions, both to test its use and to download works. I have used it only for PDF downloads, so my knowledge is limited. I certainly believe that the new download method is more accessible and usable; the old method was rather hidden, and did not have als much functionality.

[fr]

Régler le problème d'export de la page de garde de certains livres.

Solve the problem of exporting the cover page of certain books.

Neutral respondents[edit]

[en]

The use of the table of contents should be customizable and should reflect either the table of contents from the book itself or the MediaWiki's table of contents from the headings.

[pl]

Dosyć długo już funkcjonuje narzędzie, ale mimo tego po kliknięciu w "Looking for a different format?" otwiera się strona, na której nie ma słowa po polsku. Zapomniano o tym? Może dałoby się to przetłumaczyć?

The tool has been functioning for quite a long time, but after clicking on "Looking for a different format?" a page opens without a word in Polish. Has it been forgotten? Maybe it could be translated?

Not satisfied respondents[edit]

[en]

There is still need to improvement. as per Bengali Wikisource, it is not working properly.

[en]

I tried using it for the book I care about, and it didn't produce the necessary result. I couldn't find documentation for it.

Respondents who hadn't used the tool[edit]

[en]

Enable it to detect subpages automatically without requiring them to be linked in a TOC from the mainpage.

Wikimedia OCR[edit]

For more details read: Community Tech/OCR Improvements[1]

Response No. of respondents Chart
Satisfied 31
Wikimedia OCR - Wikisource Satisfaction Survey 2020.svg
Neutral 5
Not satisfied 10
Haven't used 23


Satisfied respondents[edit]

[en]

There should be instructions to improve local Tessaract only then any opensource project will be sound in future.

[fr]

Placer le bouton de façon moins gênante sur le fac similé.

Place the button in a less annoying way on the facsimile.

Neutral respondents[edit]

[en]

The "Extract Text" button was basically attached to the page; text was obscured from the tool. Making the button moveable or hideable would be better.

Not satisfied respondents[edit]

[en]

To be honest, in this era of artificial intelligence, I'm disappointed that OCR is only marginally better than it was in 1995. We really ought to be close to the point where 'OCR tools' can transcribe books on their own, but for some reason we are a long way from that and I don't know why.

[fr]

Le bouton OCR se trouve au dessus du fac similé. Il vaudrait mieux le placer dans la barre d'outil.

The OCR button is located above the facsimile. It would be better to place it in the toolbar.

Respondents who hadn't used the tool[edit]

[en]

Make it compatible with my language, which is written right-to-left.

Wikidata-Wikisource Integration[edit]

For more details read: Wikidata - Wikisource Integration Modules and Diff Blog Post 1 & Diff Blog Post 2

Response No. of respondents Chart
Satisfied 17
Wikisource-Wikidata Integration - Wikisource Satisfaction Survey 2020.svg
Neutral 5
Not satisfied 2
Haven't used 44

Satisfied respondents[edit]

[en]

All metadata should be managed by Wikidata. This could be perform a better integration between Wikisource of different languages.

[fr]

Il faudrait plus d'infobulles pour expliquer comment remplir les différents champs.

We need more tooltips to explain how to fill in the different fields.

Neutral respondents[edit]

[en]

Data not always feeding through from the Wikidata.

Not satisfied respondents[edit]

[fr]

Les éditions et les oeuvres ne permettent pas un passage one-clic entre wikipedia et wikisource :-(

The editions and works do not allow a one-click passage between wikipedia and wikisource :-(

Respondents who hadn't used the tool[edit]

[fr]

Je n'ai pas encore eu l'occasion de beaucoup utiliser ce système mais il a l'air très prometteur ! Pour renforcer les liens Wikidata-Wikisource, il faudrait avoir la possibilité d'annoter sémantiquement les textes, par exemple en liant les personnes ou les lieux mentionnés à leur élément Wikidata. Gros travail en perspective, tant dans le développement que dans l'annotation elle-même, mais qui ferait véritablement de Wikisource une bibliothèque numérique au sens actuel. Ce nouveau volet pourrait prendre la forme d'une nouvelle pastille de couleur de qualité de page, accessible une fois la page validée pour limiter les interférences entre la correction du texte et son annotation.

I haven't had a chance to use this system much yet, but it looks very promising! To strengthen Wikidata-Wikisource links, it would be necessary to have the possibility of semantically annotating texts, for example by linking the people or places mentioned to their Wikidata element. A lot of work in perspective, both in development and in the annotation itself, but which would truly make Wikisource a digital library in the current sense. This new component could take the form of a new page quality color chip, accessible once the page has been validated to limit interference between the correction of the text and its annotation.

[en]

Advertise it more on local wikisources and explain its use.

[en]

Give better information to communities. I didn't know about it before checking this link. Also, give some examples.

Wikisource Pagelist Widget[edit]

For more details read: Wikisource Pagelist Widget[2]

Response No. of respondents Chart
Satisfied 30
Wikisource Pagelist Widget - Wikisource Satisfaction Survey 2020.svg
Neutral 6
Not satisfied 4
Haven't used 28

Satisfied respondents[edit]

[en]

Loading time for a page is slow. I don't think it's necessary to use the same default resolution in this widget as the one used in Page. It should be much smaller, just enough to see the page numbers. It will make page loading much faster.

[fr]

Avoir une page d'aide dédié pas à pas.

Have a dedicated help page with step by step instructions.

Neutral respondents[edit]

[en]

Tried it once. It's an improvement, but I remember I had some difficulties finding the right numbers.

Not satisfied respondents[edit]

[en]

Needs a lot of improvement, e.g. make it more viewer friendly, perhaps divide pages in to groups, remove the tabular format, etc .

[fr]

Je le trouve très difficile d'utilisation. Saisir manuellement la pagelist a mieux fonctionné pour moi jusqu'à ce jour.

I find it very difficult to use. Manually entering the pagelist has worked best for me so far.

Respondents who had't used the tool[edit]

[en]

I believe that there is promise with this widget, although I have not used it myself. However, I believe that it is too easy to use, in that it can lead to the creation of index pages (or pagelists, at least) which are either incomplete or not representative of the file. However, that is more a problem with local policy than with the widget itself.

[fr]

mais les screenshots donnent envie de l'utiliser -:)

but the screenshots make you want to use it - :)

IA-Upload tool[edit]

Link to the tool: https://ia-upload.toolforge.org/

Response No. of respondents Chart
Satisfied 23
IA-Upload - Wikisource Satisfaction Survey 2020.svg
Neutral 6
Not satisfied 1
Haven't used 39


Satisfied respondents[edit]

[en]

Derive filename from the IA information, have some presets for filling in license information and a category chooser widget (as it is, 99% of books just get dumped onto commons with no categories). Reliability is better but still vulnerable to the Commons large-file upload bugs which they seem unwilling to address.

[fr]

Utilisé à seulement une occasion, mais a très bien fonctionné.

Used only once, but it worked great.

Neutral respondents[edit]

[en]

Repair of the former pipeline for .djvu files from IA, now that the Archive doesn't generate a .djvu file any more – so integrating the text and image files that it currently generates. And giving an error message if there's no .djvu file to upload!

[pl]

Chętnie widziałbym wsparcie dla bezpośredniego ładowania na Commons również z innych bibliotek. Na polskich Wikiźródłach wykorzystywane są przede wszystkim: Polona (https://polona.pl/) i oparte na systemie dlibra, np. PBC (https://pbc.gda.pl/dlibra) czy ŚBC (https://www.sbc.org.pl/dlibra), a z amerykańskich: Google Books i HathiTrust. Prawie nie korzystam z IA jako źródła książek.

I would love to see support for direct upload to Commons from other libraries as well. Polish Wikisource uses mainly: Polona (https://polona.pl/) and based on the dlibra system, e.g. PBC (https://pbc.gda.pl/dlibra) or ŚBC (https://www.sbc.org.pl/dlibra), and among the American ones, Google Books and HathiTrust. I hardly use IA as a source for books.

Not satisfied respondents[edit]

[en]

Is this question a joke?

Respondents who hadn't used the tool[edit]

[en]

I generally oppose the use of specialty uploading systems, such as the Upload Wizard. I worry about this, as well, because of the varying quality of metadata on IA. However, such protections are best left to individual discretion.

[fr]

Ne connais pas cet outil. Où est-il situé ?

Not aware. Where is that tool?

Other improvements that the Wikisource community would like to see[edit]

[pa]

ਨਵੇਂ ਵਰਤੋਂਕਾਰਾਂ ਲਈ ਵਿਕੀਸਰੋਤ ਦੇ ਮੇਨ ਪੇਜ਼ 'ਤੇ ਕੋਈ ਅਜਿਹਾ ਪੇਜ਼ ਬਣਾਉਣਾ ਚਾਹੀਦਾ ਹੈ, ਜਿਸ ਵਿਚ ਟੈਂਪਲੈਟਸ ਨਾਲ ਜਾਣੂ ਕਰਵਾਇਆ ਹੋਵੇ, ਕਿ ਕਿਹੜੀ ਟੈਂਪਲੈਟ ਕਿਸ ਲਈ ਵਰਤੀ ਜਾਂਦੀ ਹੈ। ਇਸ ਲਈ ਨਵੇਂ ਵਰਤੋਂਕਾਰਾਂ ਲਈ ਵਿਕੀਸਰੋਤ 'ਤੇ ਕੰਮ ਕਰਨਾ ਆਸਾਣ ਹੋ ਜਾਵੇਗਾ। ਅਜਿਹਾ ਕਰਨ ਨਾਲ ਨਵੇਂ ਵਰਤੋਂਕਾਰ ਵਿਕੀਸਰੋਤ ਨਾਲ ਲੰਬੇ ਸਮੇਂ ਤੱਕ ਜੁੜ ਸਕਦੇ ਹਨ।

For new users, there should be a page on the main page of Wikisource that introduces you to the templates and their purpose. This will make it easier for new users to work on Wikisource. Doing so will enable new users to remain connected with Wikisource for a longer period of time.

[fr]

Quand une nouvelle fonctionnalité est installé sur Wikisource, il manque de communication provenant de l'équipe de développement. Merci de trouver une meilleure façon de nous informer.

When a new feature is installed, and it's correction, there is a lack of communication from the Development team. Please find a better way to inform us, in simple terms.

[en]

Just like we have visual editor for Wikipedia where we don't need to know most of the commonly used syntaxes, we can have something similar for Wikisource where we don't have to remember most of the commonly used templates.

Notes[edit]

  1. New improvements have been made to the Wikimedia OCR, such as accessibility of the tool & multi-column support, after we ran the survey. The data here might not reflect the current satisfaction levels.
  2. New improvements have been made to the Wikisource Pagelist Widget, as part of Google Summer of Code 2021, after we ran the survey. The data here might not reflect the current satisfaction levels.