A Method for Recommending Resources Across Virtual Academic Communities Based on Knowledge Graph and Prompt Learning.

In the era of big data, virtual academic communities are flourishing and resources are growing explosively. As a result, heterogeneous fragmentation of resources and massive disorder have created constraining problems, which exacerbate the “knowledge island” effect among academic communities and challenge researchers to acquire knowledge effectively. To solve these problems, we propose a method for recommending resources across virtual academic communities (MRRVAC) based on knowledge graph and prompt learning. Firstly, we use the knowledge graph to link resources in different communities, which enables resources to be transferred between communities. Secondly, prompt learning is used to acquire the potential knowledge of knowledge graph. The final recommendation list of academic resources is obtained by training the prompt template with the improved P-tuning method and using it to mine the injected knowledge in the model. Finally, data experiments were conducted on the datasets of two virtual academic communities, Zhihu and ScienceNet. The results show that the average improvement over the original method in HR and NDCG is 0.296% and 0.271%, which validates the effectiveness of the method.