– Smithsonian transcriptions
– Crowdsourcing: cost efficient
– Question: how do you translate this to oral and video history?
– Video history: a lot of it is restricted, but some examples of video content:
– Manhattan Project
– Small arms history (including history of AK-47)
– Take reel-to-reel material on digital recorders
– Audio cassettes
– Sent to contractor to be transcribed
– Funds: sketchy
Challenges of oral transcriptions:
– Ensuring accuracy: accents, misheard words
– Tedious: pause, rewind, play, pause, rewind, play, repeat…
How do we accommodate for these problems?
– becomes more manageable if you break it up into segments: 30 seconds every 10 minutes
^ improves accuracy
– in a crowdsource context, divide it up by time stamps and assign people minute blocks
– recognizable people and recordings: entices people to participate
– have a notes portion: people can make notes of time stamps and confusion about language understanding
– But if there are unknown figures, contextualize them: e.g. why was this person important to xyz movement? How did this person influence xyz public figure?
– time stamps: makes it really clear where a project stopped
– tools to slow down speech
Potential tools:
– iTunes, Express Scribe, YouTube, iMovie (Mac; can alter pitch, speed, tone)
Data visualization:
– If you get demographics of digital volunteers, you can map where they are and what content they tend to be interested in
^ future donors and fundraisers can come from this
Why do you transcribe oral histories?
– Accessibility
– data mining: make it accessible to new research techniques
– social network analysis: transcribe and make conclusions about relationships
– preservation: if it’s recorded on a cassette or something that might be hard to digitize, you have the words forever
^ also an argument for keeping cassettes and similar media; notion that this is valuable oral content
Is there a way to build an online tool where once an oral history exists online, it can automatically enter a data mining program?
– there should be a way once an oral history is approved, you can send it directly through a code which submits it in a data context
– could be multistep, but you could build a single online forum for it
Failure is success: you rule out dead ends and apply it to your next effort
]]>Group Bibliographies
– annotated bibliographies
– shared bibliographies (e.g., through Zotero)
Transcriptions
– permanency
– skill-neutral
– “Make gov transparent and accountable through data, tools, policy and journalism”
– free, open-source; staff of designers social scientists, reporters, policy, developers, comms
– sunlightfoundation.com/api/community: projects to assist with!
What not to do:
– Data can be pretty, but does it SHOW anything
– No key, no introductory texts, no clear message of the data
– “Number art” pretty, but doesn’t tell you information
– No scales, labels, titles
“Data pervs”
– Notion that something is pretty, so it must “mean” something
– But this can mean that just because something is aesthetically appealing, people might not look deeply into it/question it
– “WTF Visualizations: Visualizations that make no sense” Tumblr page
“Squint test”
– Graphics that tell the story on the first glance
e.g. informative headlines, notable images like red lines indicating increase/decrease
– Use context: what are themes that people commonly understand? E.g. take numerical data and put it in the context of how it would fill a football field
^ most people have an estimate of the size of a football field; makes abstract data more accessible
Disseminating data
– Reporters can take screenshots of data graphics
– Reporting for reporters; making it more bite-sized for reporters to quickly understand information
– Social media shares
TOOLS:
Maps
– Show geospatial trends e.g. where are political fundraisers happening?
– www.mapbox.com
The R Project
– Data visualization coding tool online
– R-project.org
Tableau
– Basic coding
Developers/Coders:
– github.com/sunlightlabs
Sunlight Foundation Data Visualization Style Guide
Be open to and flexible with whatever tools are available! If it gets the job done, use it
]]>