An empirical analysis of the impacts of live chat social interactions in live streaming commerce: A topic modeling approach

The popularity of live streaming is driving the emergence of a new business model, known as live streaming commerce (LSC). The live chat service serves as a key indicator of viewer engagement in LSC shows, differentiating LSC significantly from traditional e-commerce. In this paper, we collect a rich live streaming dataset and identify two categories of social interactions behind live chat: transaction-oriented and relationship-oriented. Broadcasters and viewers in the transaction-oriented category focus on products and transactions and tend to communicate with non-emotional words in live chats. In contrast, broadcasters and viewers might treat each other as friends and are more likely to post emotional words in the relationship-oriented category. Our econometric model shows a curvilinear association between relationship-oriented social interactions and viewer purchase behaviors. The empirical analysis indicates that to achieve higher sales, LSC broadcasters should make relationship-oriented social interactions in live chat account for 58.64% of the probabilities. We discuss the implications of heterogeneous social interaction strategies across different broadcasters.