I’m a second year graduate student from Anhui University. My research interest includes Data Mining, Intelligent Education, Graph Neural Network and Large Language Model.

I am very fortunate to be advised by Prof. Xingyi Zhang and Prof. Haiping Ma from Anhui University.

I did my LLM algorithmic post internship at BOSS Zhipin CSL lab from September 2023 to June 2024, during which my supervisor was Dr. Chuan Qin.

I am seeking a PhD position. You can find my CV here: Siyu Song’s Curriculum Vitae.

Email / Wechat

Publication

1. Haiping Ma, Siyu Song, Chuan Qin, Xiaosan Yu, Limiao Zhang, Xingyi Zhang, Hengshu Zhu. “DGCD: An Adaptive Denoising GNN for Group-level Cognitive Diagnosis.” 33nd International Joint Conference on Artificial Intelligence (IJCAI 2024). (CCF-A) [Paper&Code]

Research experience

1. Adaptive Denoising GNN for Group-level Cognitive Diagnosis

   This study focuses on diagnosing the level of cognitive ability at the group level, with a particular emphasis on the sparsity and interaction noise of group interaction records. This study aims to address these two key issues.

2. Intelligent Recruitment Based on Cognitive Diagnostics

   This study applies cognitive diagnostic methods to the field of recruitment to conduct talent assessment and job recommendations in the recruitment field, as well as provide explainable recommendations.

3. Interpretable Recommendation Based on LLM

   This study combines the text generation capabilities of Large Language Models (LLMs) with traditional Recommendation Systems (RS) methods for interpretable recommendations. Traditional RS can provide richer historical information but lacks interpretability, while LLMs suffer from issues like forgetting and hallucinations. By leveraging the strengths of both approaches, the research aims to make more effective and interpretable recommendations.

4. LLM For Job Match

   This study uses LLM to infer and analyze suitable jobs or candidates from the candidate pool based on information about candidates or jobs. Finetuning (including SFT and RLHF) is performed on LLM using job matching data to improve LLM’s reasoning ability for job matching tasks. The web page and internal online use functions are also completed.

5. Skill System Construction in the Recruitment Field (Knowledge Graph)

   Establish a comprehensive vocational skill task system covering the entire industry to support talent recruitment and talent evaluation. Use the big model as a tool to fine-tune the LLM to implement the algorithm design of each module in the construction process and complete the corresponding tasks, reduce manpower consumption, and allow LLM to continuously complete system construction and knowledge graph construction through continuous learning.

Educational background

I earned my Bachelor’s degree in Computer Science and Technology from the College of Computer and Information Science at Anhui Normal University in 2021. I am currently pursuing a master’s degree in Artificial Intelligence at the Institute of Material Science and Information Technology at Anhui University.