Feedback to the Abstract

Feedback to the Abstract A central belief of the learning sciences is that learning is situated within communities of practice. This paper demonstrates a method for modeling learning within an online discourse community as two intersecting phenomena: members’ trajectories of participation and change in word meanings over time within the discourse community. These phenomena are well-defined by linguistic anthropology, but the ethnographic methods typically employed are difficult to apply to large communities over long timescales. We use the word2vec algorithm to train a language model on 12 million comments over a decade, capturing a snapshot every month. Then we use linguistic features of users’ posts to predict their future participation. We show how the discourse community’s word meanings shift over time with respect to relational axes such as gender and morality, and how these ideologies predict users’ trajectories of participation. 

 78 First sentence. Can we drop it? Maybe we need some warmup? 

Threw me for a loop. The 12 million comments 

Clarify how I’m fixing the ethnographic limitations

This is not a paper for AI in Education. Consider, LAK, CHI, Computational Social Science