Each year the Community Tech team at the Wikimedia Foundation runs the Community Wishlist Survey, an opportunity for Wikimedia contributors to propose new or different technical requests relevant to their work on the Wikis. Because the Community Tech team is still working on Wikipedia-focused requests from previous years, they decided to focus this year’s wishlist on sister projects other than Wikipedia, Wikimedia Commons and Wikidata.
This year, Wikisource is proving to be very prominent in the requests, with 27 of the 73 proposals (as of November 6) focused on Wikisource. We wanted to highlight a few of the proposals that reflect the kinds of needs being identified in SGill’s research around Wikisource. To see the full list, check out: the Wikisource part of the wishlist
OCR improvements focused on making it easier to support more and different languages
CIS-A2K organized Wiki Advanced Training 2019 which was focused on Wikisource for this iteration. The event had 21 participants from 9 Wikisource communities across India. SGill did a workshop with the participants to surface 5 major pain points for the Indic Wikisource community.
Based on conversations with the participants and a survey followed by that (filled by 12 out of 21 participants), the following are the top priorities around Wikisource for Indic communities:
Tools/scripts across various Wikisources - There are many scripts/tools that exist on some Wikisources and not on others. There isn’t enough documentation available to easily localize and integrate them in other Wikisources. This might also be easily implemented with Wikisource Extension.
Easy (Automated/Semi-automated) Transclusion of Proofread/Validated Indexes - Currently, the Wikisource community manually creates Main namespace pages and multiple subpages.
VisualEditor for Wikisource/Improving the Proofreading Interface - Wikisource uses a lot of templates and VisualEditor for Wikisource needs to be designed accordingly. (check interface related proposals described in the highlights above)
Bulk OCR (Not dependent on platform) - Currently, the only Bulk OCR tool available for Indic Wikisources is OCR4Wikisource which is GNU/Linux based and requires the use of a command line.
Tool to integrate book viewing with pagelist creation and identity page scan errors - the workflow for creating pagelist could be improved by an integrated bookviewer for PDF/DJVU files improving the page identification experience and allowing for adding/removing errored pages from a scanned file. (similar to the proposal Index creation wizard 2020 proposal )
Similar proposals have already been submitted as proposals in the Community Wishlist Survey 2020. This list aims to help the global Wikisource community in adding new proposals, possibly improving related proposals and hopefully prioritizing these once voting begins on 20th November 2019.