Unsupervised embeddings disentanglement using a gan for merchant recommendations

Abstract

A computer-implemented method for providing merchant recommendations comprises receiving, by a processor, raw merchant embeddings and raw user embeddings generated from payment transaction records, wherein the raw merchant embeddings include a plurality of embedded features. A generative adversarial network (GAN) performs a disentanglement process on the raw merchant embeddings to remove an effect of an identified feature by generating modified merchant embeddings that are free of the identified feature and are aligned with other ones of the plurality of features. A list of merchant rankings is automatically generates based on the modified merchant embeddings, past preferences of a target user using the raw merchant embeddings, and a current location in which the merchant recommendations should be made. A list of merchant rankings is then provided to the target user.