Quantifying the Offline Interactions between Hosts and Guests of Airbnb

In this paper, the offline interactions between hosts and guests of Airbnb are investigated. While the platform-supported communications between hosts and guests are easily tracked, new solutions are required to quantify the offline interactions. These interactions were investigated through the development of an IT artifact that determines if a review written by a guest includes a mention of a host. Manual labeling of 1,024 randomly selected reviews indicated that 85% of reviews include a reference to a host. Two primary patterns in which hosts are mentioned were discovered. A new method to detect if a host is referenced in a review is proposed. The method is based on automatically detecting these patterns using Word Embeddings and Named Entity Recognition. The method achieved an accuracy score of 91.5% and was applied on thousands of reviews from Airbnb. Results demonstrated that over 80% of reviews include references to hosts.