Empower the Graph Learning Algorithms with Semantics features from Large Language Models
Zhao Jieyi
UID: 3035844390
Department of Computer Science, The University of Hong Kong
Supervised by Prof. Huang Chao
This project aims to investigate a practical way of combining the Large Language Models(LLM) with Graph Learning. To better utilize
the summarizing and reasoning ability of LLM, the project uses LLMs to retrieve temporal and semantic features hidden in the past
interactions and descriptions. The feature obtained will be utilized in Sequential Recommendation System Models, which is a major field
that utilize Graph Learning. The final outcome of the project is a model-agnostic framework that improve the performance of
sequential recommender by optimize the representation.
The framework is composed of mainly:
