Grants:IEG/Alt text tools
This project is funded by an Individual Engagement Grant
Project idea[edit]
What is the problem you're trying to solve?[edit]
The lack of alternative image text for readers who need the alternative text to provide non-visual interpretation of images.
Text alternatives are methods of making visual information accessible through visual, auditory, or tactile modality to match the user's needs. Despite significant awareness, template changes, software changes, and a push to add to the Featured Articles criteria six years ago; only two percent of images have text alternatives on the English Wikipedia. This is failing Wikimedia's non-discrimination policy and reduces reach.
For example, with images, a partially vision-impaired person may see little or nothing of value in an image without alternative text to describe it, whereas a wholly non-visual user will only benefit from the image caption, if the image contains one. Alternative text, using the alt parameter allows non-visual users to enjoy text only descriptions of the image. With alternative image descriptions, a screen-reader can then read the alt text as well as the caption text. An alt text description is complementary to an image caption and together should provide non-visual users with a more complete emotional word picture. This puts all users of Wikipedia on a more level playing field. Ultimately, it increases their ability to assimilate an article.
For another example, some web browser setups allow alt text to be displayed on mouseover which allows sighted users with cognitive issues to more easily assimilate an image they are looking at.
| Wikipedia | Views/Hr | Size | Articles | Image-less pages | Images | Alt text |
|---|---|---|---|---|---|---|
| English (en) | 6,561,878 | 10.8 GB | 4,787,206 | 56% | 5,344,778 | 1.70% |
| German (de) | 848,184 | 3.5 GB | 1,777,086 | 54% | 2,938,529 | 0.13% |
| Japanese (ja) | 838,101 | 2.0 GB | 945,936 | 66% | 1,029,386 | 0.25% |
| Spanish (es) | 796,411 | 2.0 GB | 1,115,193 | 47% | 1,886,752 | 0.10% |
| Russian (ru) | 793,983 | 2.3 GB | 1,192,669 | 52% | 1,698,936 | 0.21% |
| Chinese (zh) | 628,241 | 1.1 GB | 797,491 | 66% | 811,159 | 0.35% |
| French (fr) | 567,021 | 2.8 GB | 1,585,562 | 54% | 2,379,061 | 1.94% |
| Italian (it) | 314,995 | 1.9 GB | 1,175,512 | 53% | 2,004,815 | 0.09% |
| Portuguese (pt) | 290,880 | 1.1 GB | 863,286 | 53% | 1,229,948 | 0.17% |
| Polish (pl) | 258,567 | 1.3 GB | 1,086,550 | 56% | 1,457,302 | 0.04% |
What is your solution?[edit]
The prototype was created at the request of the Accessibility Edit-a-Thon in Washington DC. Do to the positive outcome of that event, we'd like the opportunity to build-out and refine the tool and accompanying documentation.
We hope to create a collection of tools that allow volunteers to efficiently add text alternatives, and track changes in the model of the successful Disambiguation pages with links and Dab solver tool project. Our hope is that by creating, essentially a scored game, users will become more engaged with providing a much needed accessibility feature across the Wikipedia ecosystem. Furthermore, by introducing community features to the tool such as a leaderboard, and opportunities to connect to accessibility experts for knowledge sharing, the overall community engagement with the tool will be increased.
Project goals[edit]
We will create a tool and process to make it easier, more enjoyable, and more engaging for users to input text alternatives for images across Wikipedia.
Project plan[edit]
Activities[edit]
The activities surrounding this grant are as follows:
- Creation of a tool that will enable users to easily and enjoyably add text alternatives to images
- Build out of the "game" section of the tool by creating a leaderboard to foster friendly competition and engagment
- Development of thourough documentation surrounding the use of the tool, and best practices for providing text alternatives for images
- Hosting edit-a-thons, and meetups to foster community engagement and commitment to prolonged engagement with the tool to further accessibility
Budget[edit]
| Cost Unit | Amount |
|---|---|
| Software Development | $6,000 USD |
| Visual Design (UX and UI) | $2,000 USD |
| Documentation (well crafted and detailed) | $2,000 USD |
| Testing (to ensure accessibility of tool) | $2,000 USD |
| Travel (to support edit-a-thons and meetups) | $500 USD |
| Volunteer support (meeting spaces, refreshments, outreach) | $1500 USD |
| Total Budget | $14,000 USD |
Community engagement[edit]
We will provide WikiProjects with monthly statistics of text alternatives in their project scope and publish several articles in the Signpost on text alternatives and tool usage. Additionally, we will create barnstars to award to users and projects for their contributions in improving text alternatives. We also will look into getting Alt Text Explorer gadget-ized or enabled by default in the reading interface to expose the issue to a wider audience.
Additionally, by hosting several (no less than three edit-a-thons over the grant cycle) we hope to connect with the community and foster the incubation and speed of the adoption of the tool as a way to increase the overall accessibility of Wikipedia.
Sustainability[edit]
After the grant ends, we expect the project to be self-sustaining with user translations trickling in for other languages. Additionally, throughout this grant cycle, we will continue to look for opportunities to increase user participation, and make the tool better in future versions.
Measures of success[edit]
There are a few measurable targets:
- Using Dispenser's least popular tool (Dabfix) as a barometer, we can expect 60 more articles per month to have text alternatives added for their images. We hope for many more as image viewing is a popular activity, and we believe that the competitive nature of the game aspect will draw in additional users.
- We would also like to see prolonged adoption from the community. We plan to host no fewer than three edit-a-thons and meetups, centering the usage of the tool. Our hope is that by facilitating communication and activity among a group of editors will encourage wider-spread adoption
- Lastly, we'd like to see the 21 WikiProjects with a missing alt text percentage of over 99 reduced three points per project. This would mean the completion of alt tags for approximately 18,270 images.
FAQ[edit]
How long to complete English Wikipedia?
- We have 5.3 million image uses which need text alternatives. Here are some estimates:
- To do within five years: That's 2,929 images per day. English Wikipedia averaged 150,992 edits a day in 2015. So approximately two percent.
- To do within 15 years (Wikipedia's age): 976 images per day. Still unrealistic.
- The DPL project fix at its peak was probably fixing 1,000 links per day. So 18 years at that rate
- Dabfix, our least popular tool, which automates 70 percent of disambiguation work is used on average twice per day. It would take 7,321 years.
Can we automate it?
- Ask Dispenser again when we have automatic image categorization and an article writing bots (and rambot doesn't count). In any case, current machine learning research is focused on using a large corpus, often Wikipedia. We have not created enough good text alternatives for anyone to start yet.
Can we just have default alt text?
- Please read the current guidelines. Default text alternative came about with editors attempting to comply with the reliable sources guidelines by attempting to describe images objectively. This provided text alternatives that did not consider the context of the caption or article.
What else...?
- We limited ourselves to an available Wikipedian who was blind from birth and used the XKCD's braille "press the 2-inch button" vs "press the red button" example. This caused people to exclude color, facial expression, and other visual describers.
- Here is a fantastic example: [1] A complete accident, but it actually compares a hand built industrial burner and duplicator to plastic injected consumer burner.
- It can be fun, fitting subject knowledge and a thousand words into a tweet sized description
Other
- Wikipedia Zero has a no image option
- Screenshots
- Alt Edit
Get involved[edit]
Participants[edit]
- Full-stack Developer - Dispenser is an online tool developer with over 5 years experience building interactive tools which receive 400+ daily edits. He has participated extensively on English Wikipedia, contributed to the Featured Article process and developed Altviewer and Alt text explorer. — User:Dispenser
- Full-stack Developer - Art Frick is an online accessibility specialist and web developer. He works for the the University of New Hampshire's Institute on Disability, helping to make the web a more accessible and inclusive place. — User:Arthur.frick
- Volunteer - Whatever needs doing. I am the former founding manager of a store that specialized in specifying, customizing, selling, and implementing technology solutions and consultations to persons with disabilities all around the globe. I also am an old school web designer and web host very familiar with the need for and power of alternative images (alt img tags). Checkingfax (talk) 07:38, 29 April 2016 (UTC)
Community Notification[edit]
Endorsements[edit]
Do you think this project should be selected for an Individual Engagement Grant? Please add your name and rationale for endorsing this project in the list below. (Other constructive feedback is welcome on the talk page of this proposal).
- Support - as soon as it is ready, I am available for the use/test/export on itWiki. I help to organize the monthly "quality festivals", sometimes we introduce new tools to the community.--Alexmar983 (talk) 00:38, 6 October 2015 (UTC)
- Support - I am lousy at writing good alt image text but as a disability product implementer and disability issue advocate/activist, I strongly support alt image text implementation efforts. Alt image text is hard to write because there is no point in repeating content that is already in the article title or in the image caption: The alt text for the image has to stand on its own.
This tool created by Dispenser allows one to quickly look at images, and add alt text if they are able. This tool will encourage me to write more alt image text than I currently do. Suggestion: Add the ground rules for creating alt text right there in the tool.
Maybe alt text ideas should also be added at The Commons level when the images are uploaded or viewed.
In conclusion: I support issuing a grant to Dispenser and his team. Cheers! Checkingfax (talk) 07:19, 26 February 2016 (UTC) - Support - there is signficant need for lightweight ways for more communities to engage with Wikipedia. In the short term, we really need greater interfaces/tools that encourage participation in small ways. That being said: there is some larger community questions about how the Alttext should be stored, via Checkingfax. The advent of Wikidata and the possibility of structured data on Commons, could greatly increase the usability of these features (one alt-text description for all instances of the image. Sadads (talk) 10:49, 2016 March 1 (UTC)
- Support as the screen reader user and the blind Wikipedian mentioned above. I support anything that will increase awareness about alt text and make it easier to add, both of which are done by this proposal. Graham87 (talk) 07:14, 1 May 2016 (UTC)
- Support This is an excellent project, and I would be glad to help write alt image text. Corinne (talk) 13:53, 1 May 2016 (UTC)
- Support This looks like a fantastic endeavor, and I strongly support the implementation of alt image text. Alt image text can indeed be awkward to create while avoiding repeating the obvious, that is, content which already exists in article title or image caption. This proposal will certainly make alt text easier to add, and give us a tool which is sorely needed at this point. As a multilingual, I would love to see this great potential made available in other languages as well. Natalie.Desautels (talk) 18:54, 1 May 2016 (UTC) (User:Natalie.Desautels - for some reason the signature link is broken here, so I signed manually)
- Community member: add your name and rationale here.