Online to offline (O2O) service recommendation method based on multi-dimensional similarity measurement

With the rapid development of information technology, consumers are able to search for and buy services or products online, and then consume them in an offline store. This emerging ecommerce model is called online to offline (O2O) service, which has attracted business and academic attention. The large number of O2O services on the Internet creates a scalability problem, creating massive but highly sparse matrices relating customers to items purchased. In this paper, we proposed a novel O2O service recommendation method based on multi-dimensional similarity measurements. This approach encompasses three similarity measures: collaborative similarity, preference similarity and trajectory similarity. Experimental results show that a combination of multiple similarity measures performs better than any one single similarity measure. We also find that trajectory similarity performs better than the rating-based similarity metrics (collaborative similarity and preference similarity) in sparse matrices.