Closed forclh closed 1 month ago
请问第三种信息,在代码中是如何体现的
简而言,共性信息主要是由LLM直接总结提取,即是罗列自身节点和邻居节点的信息后,让LLM总结他们的共同特性。如在arxiv数据集中,prompt输入dst_prompt = "\n Please summarize the topic and content of these papers. All answers are in English and No Chinese in your answer",而针对电商任务,也是类似去总结用户的共同特性。 这点可以参考我们的代码里: src_prompt = '\n please summarize this paper and list five key words of this paper. All answers are in English and No Chinese in your answer'