E-mail List Archives

Search E-mail List Archives for

Number of posts in this thread: 10 (In chronological order)

From: Will Anderson
Date: Tue, Jun 18 2013 9:07AM
Subject: Making graphs built from large datasets accessible - help requested
No previous message | Next message →

Good morning everyone,
My organization is building analytics software for government child
protective services to give day-by-day insight into the current level of
care being offered the children and families under the supervision of the
agency.

While we have a mandate to be 508 compliant, we also want to do the right
thing and make the feature set accessible to all. The more people that can
understand these numbers, the better it furthers our non-profit mission.

Our data set makes it hard to build accessible charts and we're not sure if
providing the raw data would be useful . Here are some of our challenges:

- # Data points: Some of our most useful graphs are built off roughly
1,000 data points. That number is projected to increase as time passes and
we get more data in our system.
- Dynamically generated: Our graphs change based on user data entry.
Right now, they change overnight. We're planning for the graphs to be
redrawn as close-to-realtime as possible though.
- User generated: We're building tools that allow our user base to
create their own graphs meaning we can't caption them ourselves.
- Our data often show long term trends with significant short term
variation: If we provided a table, how can we help the assistive tech user
see the forrest for the trees?

We've looked for resources on the web for this problem but the closest
resource we found is IBM's Accessible Analytics: Complex Charts, Large
Datasets, and Node
Diagrams<http://www-03.ibm.com/able/news/downloads/IBM_Accessible_Analytics_CSUN_2011.pdf>;
but
that explicitly calls out many of our issues in it's "next hard problems"
slide.

Would anyone have any advice, guidance, or resources that might help us out?

Best regards,
Will Anderson

Product Manager @ Case Commons

From: John Hicks
Date: Tue, Jun 18 2013 9:35AM
Subject: Re: Making graphs built from large datasets accessible - help requested
← Previous message | Next message →

The solution to this is "sonification" surely.

Sonification has been around for a long time, but in terms of web
accessibility it has not been really exploited.

I'd love to be involved in something like that . In fact, I almost feel
silly posting to the list instead of mailing you privately and proposing we
start a company... but it is not new stuff, it is just the application of
it that is new (and maybe not even).

What is sonification? It is creating aural "soundscapes" of visual data.


It would take some getting used to, but would be a great norm if it caught
on (also providing mouse-driven extra dimensions on 2d graphics viewed by
non visually impaired people)... thinking out loud here. ...

http://en.wikipedia.org/wiki/Sonification

Interesting, let's go further!

Who's in?

best wishes all,

John


On Tue, 18 Jun 2013 11:07:12 -0400, Will Anderson < = EMAIL ADDRESS REMOVED = >
wrote:
> Good morning everyone,
> My organization is building analytics software for government child
> protective services to give day-by-day insight into the current level of
> care being offered the children and families under the supervision of
the
> agency.
>
> While we have a mandate to be 508 compliant, we also want to do the
right
> thing and make the feature set accessible to all. The more people that
can
> understand these numbers, the better it furthers our non-profit mission.
>
> Our data set makes it hard to build accessible charts and we're not sure
if
> providing the raw data would be useful . Here are some of our
challenges:
>
> - # Data points: Some of our most useful graphs are built off roughly
> 1,000 data points. That number is projected to increase as time
passes
> and
> we get more data in our system.
> - Dynamically generated: Our graphs change based on user data entry.
> Right now, they change overnight. We're planning for the graphs to be
> redrawn as close-to-realtime as possible though.
> - User generated: We're building tools that allow our user base to
> create their own graphs meaning we can't caption them ourselves.
> - Our data often show long term trends with significant short term
> variation: If we provided a table, how can we help the assistive tech
> user
> see the forrest for the trees?
>
> We've looked for resources on the web for this problem but the closest
> resource we found is IBM's Accessible Analytics: Complex Charts, Large
> Datasets, and Node
>
Diagrams<http://www-03.ibm.com/able/news/downloads/IBM_Accessible_Analytics_CSUN_2011.pdf>;
> but
> that explicitly calls out many of our issues in it's "next hard
problems"
> slide.
>
> Would anyone have any advice, guidance, or resources that might help us
> out?
>
> Best regards,
> Will Anderson
>
> Product Manager @ Case Commons
> > >

From: McMorland, Gabriel
Date: Tue, Jun 18 2013 9:42AM
Subject: Re: Making graphs built from large datasets accessible - helprequested
← Previous message | Next message →

As a blind person, I'm very interested in these problems of communicating complex ideas between the blind and sighted worlds. Unfortunately, I don't have a solution.

What do users need to do with the information displayed in the data visualization?

How do successful blind professionals navigate complex information or big data sets? There must be a blind scientist, financial analyst, or statistician who has adapted their own techniques for communicating about big data.

From: Will Anderson
Date: Tue, Jun 18 2013 9:58AM
Subject: Re: Making graphs built from large datasets accessible - help requested
← Previous message | Next message →

Thanks for the quick responses John and Gabriel. If anyone else has ideas
or resources, please send them on!

John, I'm reading up on sonification right now. It sounds really really
interesting. Would you happen to know any examples where someone has
attempted to 'sonify' a data set of graph? Please feel free to email me
directly off the thread too.

Gabriel, our non-profit was created to build more appropriate technology
for government human services workers so that they could make better policy
decisions and individual family interventions based on data. It's hard to
say exactly what actions a child welfare administrator would take given a
trend as best practice in the child welfare domain is always changing and
always dependent on the context of the local jurisdiction, agency
capability, involved employees, and particularly the family involved.

One hypothetical action an administrator could take, for instance, if they
looked at a chart that showed the number of children in foster care staying
longer than 2 years going up would be to have a conversation with the
social workers working with those children and families to understand if
the child really should be in foster care or if it's time to move toward
either reunification or adoption. Staying in foster care is often the
suboptimal result for the child and seeing that child's experience
comparatively can be a powerful motivator. Again, that's a hypothetical.

In reality, there are dozens of potential useful metrics that should be
referenced in particular circumstances. We're just starting out so we can't
give hard examples of how they're being used to make real world decisions.

Sincere thanks for engaging the question and I appreciate your point about
others dealing with large datasets. Are there any scientists, financial
folks or statisticians on the list who might be able to assist?

Best regards,
Will



On Tue, Jun 18, 2013 at 11:42 AM, McMorland, Gabriel < = EMAIL ADDRESS REMOVED = > wrote:

> As a blind person, I'm very interested in these problems of communicating
> complex ideas between the blind and sighted worlds. Unfortunately, I don't
> have a solution.
>
> What do users need to do with the information displayed in the data
> visualization?
>
> How do successful blind professionals navigate complex information or big
> data sets? There must be a blind scientist, financial analyst, or
> statistician who has adapted their own techniques for communicating about
> big data.
>
>
>
>

From: Catherine Roy
Date: Tue, Jun 18 2013 10:17AM
Subject: Re: Making graphs built from large datasets accessible - help requested
← Previous message | Next message →

Hi John,

This is a great idea. As someone who works a lot with the open data
community in Montreal, I have been struggling to offer solutions for
this type of issue. This just might be the type of proposal that could
help them get how accessibility for the blind can be part of their
projects. Plus it has the added bonus of being cool. I hope you keep
this discussion on list or at least somewhere public so we can all
benefit from this discussion.

Best regards,


Catherine

--
Catherine Roy
http://www.catherine-roy.net




On 18/06/2013 11:35 AM, John Hicks wrote:
> The solution to this is "sonification" surely.
>
> Sonification has been around for a long time, but in terms of web
> accessibility it has not been really exploited.
>
> I'd love to be involved in something like that . In fact, I almost feel
> silly posting to the list instead of mailing you privately and proposing we
> start a company... but it is not new stuff, it is just the application of
> it that is new (and maybe not even).
>
> What is sonification? It is creating aural "soundscapes" of visual data.
>
>
> It would take some getting used to, but would be a great norm if it caught
> on (also providing mouse-driven extra dimensions on 2d graphics viewed by
> non visually impaired people)... thinking out loud here. ...
>
> http://en.wikipedia.org/wiki/Sonification
>
> Interesting, let's go further!
>
> Who's in?
>
> best wishes all,
>
> John
>
>
> On Tue, 18 Jun 2013 11:07:12 -0400, Will Anderson < = EMAIL ADDRESS REMOVED = >
> wrote:
>> Good morning everyone,
>> My organization is building analytics software for government child
>> protective services to give day-by-day insight into the current level of
>> care being offered the children and families under the supervision of
> the
>> agency.
>>
>> While we have a mandate to be 508 compliant, we also want to do the
> right
>> thing and make the feature set accessible to all. The more people that
> can
>> understand these numbers, the better it furthers our non-profit mission.
>>
>> Our data set makes it hard to build accessible charts and we're not sure
> if
>> providing the raw data would be useful . Here are some of our
> challenges:
>> - # Data points: Some of our most useful graphs are built off roughly
>> 1,000 data points. That number is projected to increase as time
> passes
>> and
>> we get more data in our system.
>> - Dynamically generated: Our graphs change based on user data entry.
>> Right now, they change overnight. We're planning for the graphs to be
>> redrawn as close-to-realtime as possible though.
>> - User generated: We're building tools that allow our user base to
>> create their own graphs meaning we can't caption them ourselves.
>> - Our data often show long term trends with significant short term
>> variation: If we provided a table, how can we help the assistive tech
>> user
>> see the forrest for the trees?
>>
>> We've looked for resources on the web for this problem but the closest
>> resource we found is IBM's Accessible Analytics: Complex Charts, Large
>> Datasets, and Node
>>
> Diagrams<http://www-03.ibm.com/able/news/downloads/IBM_Accessible_Analytics_CSUN_2011.pdf>;
>> but
>> that explicitly calls out many of our issues in it's "next hard
> problems"
>> slide.
>>
>> Would anyone have any advice, guidance, or resources that might help us
>> out?
>>
>> Best regards,
>> Will Anderson
>>
>> Product Manager @ Case Commons
>> >> >> > > >

From: John Hicks
Date: Tue, Jun 18 2013 10:25AM
Subject: Re: Making graphs built from large datasets accessible - help requested
← Previous message | Next message →

Yes, I think these sorts of things work best in the open!

And it is a very cool subject.

Perhaps over the next days a number of folks might identify themselves as
interested and we can look more closely at it.

I thought a lot about it 3 years ago. One of the problems was the sound,
because Flash was the way... but now that is changing with the HTML 5 sound
embedding features.

I already had another enthusiastic response. This was a great question
Will raised, inspirational.

Let's share more very soon,

thanks

John



On Tue, 18 Jun 2013 12:17:26 -0400, Catherine Roy
< = EMAIL ADDRESS REMOVED = > wrote:
> Hi John,
>
> This is a great idea. As someone who works a lot with the open data
> community in Montreal, I have been struggling to offer solutions for
> this type of issue. This just might be the type of proposal that could
> help them get how accessibility for the blind can be part of their
> projects. Plus it has the added bonus of being cool. I hope you keep
> this discussion on list or at least somewhere public so we can all
> benefit from this discussion.
>
> Best regards,
>
>
> Catherine

From: McMorland, Gabriel
Date: Tue, Jun 18 2013 10:26AM
Subject: Re: Making graphs built from large datasets accessible - help requested
← Previous message | Next message →

I'm thinking about how to show both the forest and the trees to a blind decision maker. I think I'd want to get a feel for data trend using the sonarscape, and then make specific queries about data. Simple example: I want to listen to the stock market and also get a precise reading of the dow jones average for any point in the "data song".Yes, I also hope to see this conversation on the web aim list. I'm a blind person that went to school for urban planning and public policy, so I know firsthand how important it is to have access to data that informs strategic decisions.
It is a real barrier to my professional options if I can't access data trends that inform policy analysis and management strategy.

From: Chagnon | PubCom
Date: Tue, Jun 18 2013 11:28PM
Subject: Re: Making graphs built from large datasets accessible - helprequested
← Previous message | Next message →

Years ago I worked with a blind statistician who worked on a gigantic 4000
cell table.
One strategy that worked in this situation was to divide up the huge Excel
spreadsheet into smaller chunks of data in linked worksheets (or tabs).

The final spreadsheet consisted of:
Tab 1, a table of contents that linked to the other tabs.
Tab 2, a summary of the individual worksheets/tabs with totals compiling the
data from all the worksheets/tabs.
Tabs 3, 4, etc. were smaller tables by topic. Tab 3 was blue widgets, Tab 4
was red widgets, etc.

Not only did it make it easier for him to navigate and use the spreadsheet,
but sighted users benefitted, too.

-Bevi Chagnon
- - -
www.PubCom.com - Trainers, Consultants, Designers, Developers.
Print, Web, Acrobat, XML, eBooks, and U.S. Federal Section 508
Accessibility.
New Sec. 508 Workshop & EPUBs Tour in 2013 - www.Workshop.Pubcom.com

From: Will Anderson
Date: Wed, Jun 19 2013 9:08AM
Subject: Re: Making graphs built from large datasets accessible - help requested
← Previous message | Next message →

Bevi, would you happen to have the contact info for your former
statistician colleague? Seeing that spreadsheet would probably be super
helpful for our team.

Thanks!

-Will


On Wed, Jun 19, 2013 at 1:28 AM, Chagnon | PubCom < = EMAIL ADDRESS REMOVED = >wrote:

> Years ago I worked with a blind statistician who worked on a gigantic 4000
> cell table.
> One strategy that worked in this situation was to divide up the huge Excel
> spreadsheet into smaller chunks of data in linked worksheets (or tabs).
>
> The final spreadsheet consisted of:
> Tab 1, a table of contents that linked to the other tabs.
> Tab 2, a summary of the individual worksheets/tabs with totals compiling
> the
> data from all the worksheets/tabs.
> Tabs 3, 4, etc. were smaller tables by topic. Tab 3 was blue widgets, Tab 4
> was red widgets, etc.
>
> Not only did it make it easier for him to navigate and use the spreadsheet,
> but sighted users benefitted, too.
>
> -Bevi Chagnon
> - - -
> www.PubCom.com - Trainers, Consultants, Designers, Developers.
> Print, Web, Acrobat, XML, eBooks, and U.S. Federal Section 508
> Accessibility.
> New Sec. 508 Workshop & EPUBs Tour in 2013 - www.Workshop.Pubcom.com
>
>

From: McMorland, Gabriel
Date: Wed, Jun 19 2013 10:10AM
Subject: Re: Making graphs built from large datasets accessible - help requested
← Previous message | No next message

Will,

You may want to contact the WGBH National Center for Accessible Media, "A non-profit R&D organization dedicated to achieving media access equality for people with disabilities.". They don't have anything about big data published on their website, but this problem seems like something they would have useful insight on. Also, it might bring their research team into the conversation.

http://ncam.wgbh.org/

Archimedes project at Stanford, researching new interfaces for disabled users.
http://archimedes.hawaii.edu/

Also, National Federation of the Blind
www.nfb.org

American council of the blind
www.acb.org