Qichuan Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Qichuan Liu has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
- Chenfeng Zheng
Xiamen University, Xiamen, China
1 shared publication - Chentao Zhang
Xiamen University, Xiamen, China
1 shared publication - Gefei Zhang
School of Informatics, Xiamen University, China
1 shared publication - Hong Zhao
State Grid Gansu Electric Power Research Institute, China
1 shared publication - Jinxiong Zhao
State Grid Gansu Electric Power Research Institute, China
1 shared publication - Qinggang Zhang
The Hong Kong Polytechnic University, Hong Kong, Hong Kong
1 shared publication
Latest Publications
- Open Access
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 2296-2307
Facilitating Generative Retrieval with Logical Denoising for Interpretable Conversational Search
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 2296-2307
Open AccessFacilitating Generative Retrieval with Logical Denoising for Interpretable Conversational Search
- Open Access
- 25 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Power Large Language Model Exploration: Activation, Measurement and Enhancement for Operations and Maintenance Knowledge: Activation, Measurement and Enhancement for Power O&M Knowledge
With the rapid advancement of Large Language Models, their applications are gradually transitioning from general to specific domains. However, the application of LLM in the electric power domain is still in its early stages, and few studies have explored power LLM. Currently, there are two main challenges against power LLMs: (1) determining how to measure the real power knowledge capacity of LLMs to facilitate targeted enhancement of specific knowledge. (2) identifying practical enhancement methods to facilitate efficient and feasible power LLM applications in real-world scenarios. In this paper, we ask three insightful questions that address the power knowledge capacity of LLMs and then draw inspiration from Reflexion and CoT to design an Activation, Measurement and Enhancement framework (AME) for power operations and maintenance (O&M) knowledge. Specifically, we ask three “HOW” questions based on the activation, measurement, and enhancement of power O&M knowledge. We introduce a Reflexion Module to discover the knowledge capacity of LLM and a Knowledge Graph Module to provide external knowledge of LLM in our proposed AME. Experiments on the real-world dataset provide strong evidence when we answer the above three insightful questions.
- 25 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessPower Large Language Model Exploration: Activation, Measurement and Enhancement for Operations and Maintenance Knowledge: Activation, Measurement and Enhancement for Power O&M Knowledge
With the rapid advancement of Large Language Models, their applications are gradually transitioning from general to specific domains. However, the application of LLM in the electric power domain is still in its early stages, and few studies have explored power LLM. Currently, there are two main challenges against power LLMs: (1) determining how to measure the real power knowledge capacity of LLMs to facilitate targeted enhancement of specific knowledge. (2) identifying practical enhancement methods to facilitate efficient and feasible power LLM applications in real-world scenarios. In this paper, we ask three insightful questions that address the power knowledge capacity of LLMs and then draw inspiration from Reflexion and CoT to design an Activation, Measurement and Enhancement framework (AME) for power operations and maintenance (O&M) knowledge. Specifically, we ask three “HOW” questions based on the activation, measurement, and enhancement of power O&M knowledge. We introduce a Reflexion Module to discover the knowledge capacity of LLM and a Knowledge Graph Module to provide external knowledge of LLM in our proposed AME. Experiments on the real-world dataset provide strong evidence when we answer the above three insightful questions.