Crowdsourcing without internet
- How to include the knowledge of people without Internet access in our databases?
Our Solution: Interactive Voice response
Instead of relying on the Internet to interact with communities that have expert knowledge in a specific field, we propose a voice-based interface using the telecommunication network.
The number of mobile connections observed by the Digital 2022 report on Ghana suggests that the large majority of the population is actively using the telecommunication network, even in rural communities.
About the Project
The aim of this project is to use crowdsourcing techniques to tag objects from the colonial period exhibited in Dutch museums. We created an IVR (Interactive Voice Response) system to gather information about colonial heritage objects that are originally from Ghana, and that are now exhibited in the Wereldculturen Collectie. We are using the Twilio platfrom, which offers a Voice API to create a personalized voice interface.
The IVR system developed in this project allows participants to answer questions about selected objects through a voice interface. The goal is to extract information about these objects and create a polyvocal dataset.
This paper aims to identify and propose an effective method and approach for using nichesourcing in areas with low Internet connectivity. The primary focus of this study is to develop a robust methodology and technical implementation. While we test our voice application in the Netherlands, conducting the experiment in Ghana or another sub-Saharan country is out of the scope of this research. By prioritizing the development of a methodological framework and implementation, this research seeks to create the building blocks for performing voice-based crowdsourcing targeted at expert communities.
Research Questions
- RQ1: Is an IVR system a reliable option to extract information about colonial heritage objects displayed on printed images?
- RQ2: How can we include participants’ responses in the database to create a polyvocal representation of each object?
Our methodology and results are outlined in our research paper. You can access it here.
Selected objects
Object 1: Drum
Object 2: Kuduo Pot
Object 3: Fertility doll
Contact information
Find all the code at this Github repository.
If you have any questions or would like to get in touch, please feel free to reach out:
- Email: n.maronic@gmail.com
- LinkedIn: Naomi Maronic