So let’s say somebody gives you 100 stories about your work and you’re too busy to read them. How about a picture that shows how all the words in those stories connect to each other? Would that help you assimilate this knowledge, borg style?
Using the free python library networkx and free Gephi software visualization software, I can now build a network map of all word associations in a group of stories (selected using keyword searching). The spherical map is equivalent to this network map of words that appear next to each other in hundreds of stories that mention step parents:
The reason I chose to search for stories that include one of these phrases (step mother, step father, parent) is that RETRAK has observed that many kids appear to run away from familes and live on the streets of Kampala when one parent dies and the surviving parent remarries.
What this map reveals is that the number one theme associated with parents is school. School fees, going to school, street children, child labor, death of a parent, single parent, and paying for school all appear on this map, and RETRAK has yet to submit a single story from their population of Kampala street kids!
So I looked at 7000+ stories mentioning “school”,”fee”,”student”, or “education”. It gets a bit noisy, so I present the full picture and then some heavily filtered ones, only showing the most common phrases:
I continue to use “school fees” as my test phrase because it appears very frequently in stories but is not usually the major focus of organizations. In fact, the government of Kenya promises free universal education. But apparently school continue to find alternative ways to charge students, through books, uniforms, and lunch fees that keep many poor kids away. 7000 stories don’t lie! School fees remains a major problem, and overlaps with those word maps of step parents, street children, and lack of opportunity.
Other story theme derived from mapping word associations
Kenya Drought, “no water” stories
Famine (31 stories)
FGM, Female Genital Mutilation, Circumcision
Now this map surprised me the most. First, finding stories about FGM in the collection of 20,000+ is quite difficult because people use so many code words to talk around the issue. This map reveals some shocking overlapping issues. “Reproductive health” is the NGO phrase (and a deeper look at the story patterns should reveal stories with these phrases come from “actors” in changing the practice) But HIV/AIDS is also an important issue here. And the Kisii community in particular is very closely associated with FGM, but not “female genital circumcision.”
It’s your turn!
Not bad for an hour of analysis. Ready to try it yourself? email me (you’ll find my email on the GlobalGiving About US page) and I’ll help you. – Marc Maxson