Reading 43,388 Stories to Find Trends
GlobalGiving used BigML to find trends in 43,388 stories. What’s the story behind that?
GlobalGiving is an online marketplace that connects givers to causes and countries they feel connected to. The website features a large number of projects in various countries and with a multitude of causes. You can select any of these projects to donate to. Subsequently you’ll be kept up to date on the progress of your project.
Of course you want your donation to make a good (social) return, so measuring effectiveness is paramount. GlobalGiving has learned that one of the best ways to contribute to social change is to develop better feedback loops. So they set up the Storytelling Project: an experiment in collecting community feedback. Thousands of stories have now been collected, told by people from areas where GlobalGiving partners work. All these stories are an answer to one simple question: “Tell us about a time when a person or an organization tried to change something in your community”.
Faced with the challenge of analyzing these stories, GlobalGiving’s Marc Maxson turned to BigML. Would it be possible to find some trends in these stories? See the riveting conclusion in Marc’s blog post “Using BigML to dissect trends in 43,388 stories” (Spoiler: Yes, it was possible). Marc told us in his next round he wants to take a different approach. He will deconstruct the narratives using python’s natural language processing toolkit (nltk) into elements and running BigML on these elements to answer a more fundamental question: “What makes a good story (for evaluation purposes)?” So keep an eye on his ‘Chewychunks’ blog if you are interested!
BigML is happy to have saved GlobalGiving time and effort they would have had to spend otherwise in analyzing these stories. If you have a social project in mind, that could benefit from BigMLs predictive models please let us know and we’ll be happy to contribute.