Swara Meets Ushahidi: Integrated Voice Response Meets Mapping

Feb 27, 2012

[Cross-posted from the MojoLab, by Arjun Venkatraman]

A couple of months ago, while I was visiting Borneo, I was looking at the Swara blog interface and wondering if there was some way to make it more exciting and informative, so that even people who didn’t necessarily want to wade through audio files could still get a clear visual snapshot of the subjects being discussed.

The usual ideas like tag clouds and slide shows would do part of the trick, but what I really wanted was a geographic representation of the data, so that viewers (including us at Swara) could tell where the content was coming from and who were the people generating it. Still pretty new to the social tech space, I resorted to Google and a few searches on the lines of “map data visualization” threw up Ushahidi. I had heard the name at Hackathons and I knew that they ranked in the top ten non profits in the world, but beyond that I hadn’t a clue about what they did.

From the Ushahidi site:

“Ushahidi”, which means “testimony” in Swahili, was a website that was initially developed to map reports of violence in Kenya after the post-election fallout at the beginning of 2008. Since then, the name “Ushahidi” has come to represent the people behind the “Ushahidi Platform”. Our roots are in the collaboration of Kenyan citizen journalists during a time of crisis. The original website was used to map incidents of violence and peace efforts throughout the country based on reports submitted via the web and mobile phones. This website had 45,000 users in Kenya, and was the catalyst for us realizing there was a need for a platform based on it, which could be used by others around the world. Since early 2008 we have grown from an ad hoc group of volunteers to a focused organization. The team is comprised of individuals with a wide span of experience ranging from human rights work to software development. We have also built a strong team of volunteer developers primarily in Africa, but also Europe, South America and the U.S.

My moment of Ushahidi enlightenment came at a session by Anahi Ayala in Kabul, Afghanistan of all places, where I was attending the Kabul Innovation Lab organized by Internews in partnership with USAID. By the time Anahi had finished speaking I had a running Ushahidi installation on my laptop (at the time running Fedora 15). Minus a few configuration hacks that I had to Google out, the installation was fairly straightforward and my next step was to figure out how to plug in the Swara content. The ideal way would be to actually have the Swara IVR insert records directly into the Ushahidi database, but that would take too long to code and I WAS at a Hackathon after all.

So instead of doing the “Right Thing” I simply pointed the RSS feed from Swara into the Ushahidi News feed section and voila! I had my first Swara-Ushahidi (purely alphabetical order  ) mashup.

I’ve spent the last month or so since then simply playing around with different Ushahidi features like categories, image and other media support and the map sources section. I’ve also had an opportunity to speak with the Ushahidi folks and they’re rather keen on having an IVR interface into Ushahidi as well. The big drawbacks of the rudimentary RSS based integration are that

The content manager still has to go and convert eache RSS item into a report manually. This takes a long time and slows things down.

The location on the posts still needs to be tracked manually.

Automation can take care of both these issues. The primary Swara loop can also plug records straight into Ushahidi. Ideally I would love to use a Loudblog like interface (essentially with a flash player to preview audio files) for moderation and the Ushahidi interface to publish to. That plus an auto release feature for known callers/users would probably be an excellent place to start

For the second, I think that while a phone call doesn’t say where the call is coming from, some intelligent guesswork can still help automate the location hunting. For example, if a moderator once tags a number with a location, that location could be used as default for all calls coming from that number in the future.

A little more advanced approach could be to use the cell ID. Most phones know their cell ID, as the Cell Info Display is still a pretty common feature. If a feature phone could be programmed to include the Cell Info into SMSes sent to the Swara number by default, that could help pinpoint a location too. Initilally, some system training might be required, but I think over time, we could build a database mapping cell IDs to geolocations (lat, long) and that could help locate even people with low end phones.

That would be a killer feature to have and would enhance the effectiveness of crowdsourcing in general in a big way. We tested out the mashup during a field trip in Central India (Khandwa)

We spent the day meeting women’s Self Help Groups. These groups run a common savings program, wherein women from the villages form groups and pool a portion of their savings. The pool is then used to give loans to members of the groups as needed. They are exploring the use of the IVR as a means to record field data. For example the women’s self help groups would record reports of their weekly meetings including data such as the leaders name, group name, weekly savings etc.

This will enable cheap, accurate and transparent records of such groups, since this data can be recorded in under three minutes provided a format, which puts the cost of each report at just under INR 2 (USD .05). The reports can be released onto the same IVR to let groups share information.

We also explored using the same model to report on other activities, such as cooperative dairy projects and NREGA payment reports.

. In most cases, simply by using a custom form, Ushahidi could provide a very effective reporting tool. This could also be applicable in other geographies and for other applications. Additionally, by developing a little intelligence on the IVR side, localized services could be layered on top of this sort of system. For example, field workers entering an area can get a fairly accurate idea of the ground realities by tuning in to the closes Swara-Ushahidi mashup and view additional media from the area.

The same server could be set up to receive input from multiple channels i.e. Bluetooth, MMS and WiFi in addition to the Internet and the IVR. The next item on my task list is to add FrontlineSMS to the mix and test drive the whole contraption in the field for a month to see what we get and whether 3 platforms are actually better than one.