Digital payment apps, e.g., available via smartphones, can make multiple payment instruments available to a user; however, a user may not link all their cards to a payment app. Some payment apps include analytics features based on data from cards linked to the app. However, such data can give an incomplete financial picture of the user. This disclosure describes techniques to develop a more complete representation of a user that has provided only a partial view of their finances. Two embedding vector spaces are created, one trained over users with incomplete financial profiles and another trained over users with complete financial profiles. A map is created between the two vector spaces. The map is used to extend the representation of a user with a partial financial profile to that of a user with a complete financial profile.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Kota, Nithya; Dewan, Maneesh; Mallya, Ganesh; and Chhugani, Jatin, "Using Cross-User Understanding to Develop Better User Embeddings", Technical Disclosure Commons, (September 08, 2022)