Search

Online communities, such as social networks and peer production communities, are an essential part of our daily life. They act as an information system for various purposes and have drawn the attention of research from multiple disciplines. One thread of research studies online communities through the lens of social science theories to understand community insights for design implications. Another thread of research involves the development and support of artificial intelligence that is used to build and integrate algorithmic technologies into online communities to assist humans with various tasks. However, a series of challenges and issues have been discussed in building those intelligent systems, such as lack of understanding of community social dynamics and stakeholders’ values, and lack of transparency of trade-offs in algorithm design. In addressing these challenges, my dissertation connects these two threads of research. It takes human-centered approaches by conducting analytical studies through the lens of social science. These studies guide the building of effective intelligent algorithmic systems in online communities and inform the design of interactive interfaces that communicate algorithm trade-offs to stakeholders who lack technical backgrounds in order to support their decision making. My dissertation contributes to the ongoing conversations about AI system building for online communities at a large scale and about participatory design of machine learning systems with lay people.