In this log analysis, I apply machine learning and text mining techniques to understand what features of members' posts increase community response in Microsoft's interest sharing network, So.cl. I used C# to gather and clean the data and machine learning packages developed in Java.
This project was done as part of a summer internship I completed at FUSE Labs in Microsoft Research. In the first few of twelve weeks, I brainstormed research questions to ask about So.cl, Microsoft's interest-sharing network. It was important to me that my research questions and answers would provide some useful insight into user behaviors to both the research community and the development team. I worked with the product-oriented engineering team to retrieve the data necessary to answer my questions. Using C#, I cleaned and organized the data for my purposes and then applied discourse analysis methods to analyze the interest-sharing network data. This opportunity allowed me to experiment with a variety of automated topic modeling and clustering techniques. My findings were published in a short paper in the Proceedings of Human Factors in Computing Systems 2013 (CHI13).
As a side bonus, I was also interviewed about my experience as an MSR intern for the MSR website news.
Howley, I., & Newman, T. (2013, April). Factors impacting community response in an interest-sharing network. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 2283-2286). ACM.