Causal Discovery in Recommender Systems: a Case Study in Online Hotel Search

We approached the causal discovery task in the recommender system domain to learn a causal graph by combining observational data provided by a meta-search booking platform for online hotel search with prior knowledge made available by domain experts. The results show that it is possible to learn a causal graph coherent with previous findings in the recommender systems literature about the relations between different factors. Furthermore, we also discovered new insights that could help in the recommendation process.