Abstract
Systems and methods are described for cross-platform, social-context-aware product recommendation. Interaction signals associated with a user are aggregated across multiple applications and normalized into features used to populate an inspiration graph representing entities and relationships reflecting user interests and context. A conversational shopping mode receives multi-turn requests and constraints and may accept multi-modal inputs such as room, wardrobe, or invitation images. Candidate products are retrieved from one or more catalogs and ranked using features derived from the inspiration graph and, when provided, the multi-modal inputs. In some embodiments, a trend momentum score is computed from velocity and recency of style adoption across applications and used to influence ranking. In some embodiments, gift affinity scoring is computed for a selected recipient using recipient-related social context signals subject to privacy controls. A generative visualization may render recommended products into user-provided visual contexts.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Anonymous, "Multi-Modal Cross-Platform Recommendation System with Aggregated Social Context Signals", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10647