*What*: Aditya Vashistha
*When:* Tuesday, May 20th at 12pm
*Where:* The Allen Center, CSE 203
Please join us for this weeks Change Seminar. CSE graduate student
Aditya Vashistha will be presenting two new ideas he has for future work
and looking for feedback from the Change community.
*Abstract:*
In this brainstorming session, we will talk about two projects at the
intersection of ICTD and Human Computation.
*Voice-based Microtasking Platform*
Our aim is to design, build and evaluate a voice-based microtasking
platform to enable low-income low-literate marginalized communities in
India to perform wide range of audio micro tasks using their basic
phone. The microtasking platform will run on an Interactive Voice
Response (IVR) service where people will call a number to perform tasks
from a collection of audio tasks. For tasks that require objective
response like multiple choice response, a worker will input response by
pressing keys on the phone. For tasks that require subjective response
like translation task, a worker will perform task by recording the audio
response in an active call session. Workers will be compensated for
their efforts in the form of mobile airtime top-ups which are very
common in the developing countries.
*Second Project: Generating Keywords for Audio Content in Local Languages*
In recent years, various researchers and practitioners have deployed IVR
systems to collect data from and provide information to low-literate
low-income marginalized communities. Some notable IVR systems include
CGNet Swara for citizen journalism, Avaaj Otalo for agriculture
information discussion, IBM's Spoken Web for local content creation and
dissemination, Polly and Sangeet Swara for providing entertainment
content, Healthline for disseminating health information, and Ila
Dhageyso for civic engagement. Though these IVR systems have been
successful in engaging marginalized communities in instrumental as well
as non-instrumental discussions, one of the major drawback of these
systems is the inability to search and index the content generated in
languages spoken in the developing world. This deteriorates usability
for both IVR users, and Internet users who browse these datasets.
Our solution is to create a crowdsourcing smartphone application which
will play audio files one after other. Users will choose icons
corresponding to the keywords they identify while listening to audio
files, and thus will generate keywords for the audio content produced in
local languages.
*About the speaker:*
Aditya Vashistha is currently pursuing PhD in Computer Science at the
University of Washington where he design, build and evaluate
technologies for marginalized rural and urban communities to improve
their access to information. His research focus is to build and evaluate
scalable voice based communication platforms that enables the next
billion people to access the Internet, social media platforms, and
crowdsourcing platforms on their non-Internet-enabled mobile phones.
Earlier, he also conducted research at Microsoft Research India and
Infosys Labs. You can read more about him at www.adityavashistha.com
_______________________________________________
change mailing list
change@change.washington.edu
http://changemm.cs.washington.edu/mailman/listinfo/change
