Topic Analysis and Visualisation of Peer-to-Peer Platform Data: An Airbnb Case Study

Abstract

Peer-to-peer (P2P) platforms play an important economic role as they bring together buyers and sellers and allow them to directly interact with each other. People who sell goods or services on P2P platforms are often ordinary citizens who do not have sophisticated marketing knowledge or skills. It is important that service providers are empowered to know how to market themselves since they are competing globally. In a review of accommodation P2P platforms, previous studies have shown the importance of marketer-generated content (MGC), and how it relates to aspects such as pricing, demand, and customer experience. The same holds for user-generated content (UGC), which typically takes the form of customer reviews. However, there was a lack of studies considering both host and guest generated data. This study addresses the identified gap by using topic modelling to analyse both host and guest data obtained from the Airbnb platform. A Latent Dirichlet Allocation algorithm is used to discover latent topics in the Airbnb MGC and UGC, respectively. The discovered topics are first ranked and then analysed using Tableau’s data visualisation tool, by using various dimensions such a geographic location, review scores, or number of reviews. The analysis, even among the top-ranked properties, shows that there are still many mismatches between host and guest topics. The paper contributes by illustrating the value of topic modelling and analysis to gain practical insights into the data accumulated on an accommodation booking platform.