Questions in Appendix

[YanhuiS]

非常感谢您团队的伟大工作，现有问题向您请教，请问E.2 Social Network中：

1. 80%比例的core agent数据是如何体现在网络构建中的呢，
2. 核心智能体跟常规智能体的区别是什么？核心智能体是如何构建的？
3. 兴趣一致是指两次独立采样后，两个人的两个兴趣都一样吗？

期待您的回答。

Translate:

Thank you very much for the great work of your team. I have some questions regarding E.2 Social Network, and I would like to seek your guidance:

How is the 80% proportion of core agent data reflected in the network construction?
What is the difference between core agents and regular agents? How are core agents constructed?
Does "interest alignment" mean that both interests of two people are the same after two independent samplings?

[zhangzaibin]

The core users are selected based on real user profiles. Additionally, we will give more followers to these core users, which aligns better with the distribution characteristics of social media users (a small number of users have large followers, while the majority have fewer followers). Regarding interests, we use prompts to set constraints.

[YanhuiS]

你好，新年快乐，我可能没有表达清楚我的问题，

1. 核心智能体的比例是 80 %还是 20%，这个数据主要来源于马太效应paper的那一页？核心智能体在网络信息传播过程中起到什么作用，与常规智能体有什么区别？
2. 核心智能体是如何挑选的，具体选哪些真实用户用来构建人物画像？
3. 兴趣一致是指“两次独立采样后，两个人的两个兴趣都一样”吗？您说的“Regarding interests, we use prompts to set constraints.”是什么意思？