Qingjun Cui

Title: Qingjun Cui and Graph-Based Entity Representations

Introduction

Qingjun Cui is an inventor based in Sunnyvale, California, and is associated with Amazon Technologies, Inc. His patent work explores how language models can combine information from graphs, text, and images to represent entities.

Latest Patents

Cui is associated with 1 patent: “Systems for Determining Representations of Entities Based on Graph Data Using a Language Model.” The invention describes generating separate representations from graph data, text, and images related to an entity, then providing those representations to a large language model to produce an overall entity representation. The graph data captures entity characteristics and relationships with other entities. The resulting representation can support other models in identifying entities with similar or differing characteristics.

Career Highlights

Cui’s listed patent reflects work at the intersection of graph data and language models. It describes a method for combining multiple data types into a representation that can be used in downstream model tasks.

Collaborations

The provided information lists Houyu Zhang and Seongjun Yun as Cui’s coworkers. It does not specify their roles in the patent or the nature of their collaboration.

Conclusion

Qingjun Cui’s patent focuses on using a language model to create entity representations from graph, text, and image data, with potential applications in comparing entities.