Ming Yan

Title: Innovations by Inventor Ming Yan

Introduction

Ming Yan is a notable inventor based in Xi'an, China. He has made significant contributions to the field of technology, particularly in the development of advanced recommender systems and data tagging methodologies. With a total of 3 patents to his name, his work continues to influence the industry.

Latest Patents

Ming Yan's latest patents include an innovative LLM-based recommender system for data catalogs. This system utilizes probabilistic methods to create coherent responses, addressing the challenge of 'LLM hallucination.' By incorporating asset metadata into the prompts, the system enhances the LLM's understanding of available assets. The design allows for a limited set of candidate assets to be included in the prompt template, which helps mitigate hallucination issues. Additionally, the system supports multiple rounds of interaction, fostering dynamic user engagement. The output is a recommended asset list that is both accurate and interpretable, thanks to the LLM's advanced context understanding.

Another significant patent is the data tagging and prompt generation system. This system operates by receiving input data and identifying relevant metadata. It generates statistics based on the input data and calculates a sample size accordingly. A prompt is then created based on a template and provided to a language model for tagging the input data. The output includes tagged data, which can be queried for further analysis.

Career Highlights

Ming Yan is currently employed at SAP SE, where he applies his expertise in developing cutting-edge technologies. His work at SAP SE has positioned him as a key player in the advancement of data management solutions.

Collaborations

Ming collaborates with talented individuals such as Jing Guo and Jingtao Li, contributing to a dynamic and innovative work environment.

Conclusion

Ming Yan's contributions to technology through his patents and work at SAP SE highlight his role as a leading inventor in the field. His innovative approaches to recommender systems and data tagging continue to shape the future of data management.