A new framework called the Multiplex Multimodal Representation Model (MMRM) has been developed to enhance product ranking in e-commerce search by leveraging diverse multimodal data. Unlike previous methods that use single collaborative signals, MMRM aligns with multiple signals simultaneously to generate comprehensive item representations. It also introduces a multiplex user representation strategy that models user behavior sequences using these rich item representations. This approach has been successfully deployed on the JD e-commerce search engine, leading to significant performance improvements for millions of users. AI
IMPACT This model could improve e-commerce search relevance and user experience by better utilizing multimodal product information.
RANK_REASON The cluster describes a research paper detailing a new model and its application.
Read on arXiv cs.IR (Information Retrieval) →
- alphaXiv
- arXiv
- CatalyzeX
- DagsHub
- Gotit.pub
- Hugging Face
- JD.com
- Multimodal Large Language Models
- ScienceCast
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
