Dr. Susan Herring and Fernando Maestre qualitatively coded about four hundred posts on forums for people living with HIV in an attempt to capture the relationships between how people seek support and the support that they receive. Posts were divided into “support seeking”, “support providing”, and “other”. Support seeking was further divided into “direct” and “indirect”, and “emotional” and “informational” (as well as “network” and “esteem”, but these categories were too sparse to be considered). Support providing posts were also divided between “informational” and “emotional” with personal centeredness scores of “low”, “medium”, and “high” based on how attentive the response was to the original poster’s specific concerns.
I have been working to come up with a machine learning algorithm that can accurately replicate these qualitative codes and that we can automate the data collection for more responses. With more samples, I could perform a detailed analysis of the relationship between the content and framing of a post and the type of support it receives. Our initial findings were published in Information in 2018.