WES: Agent-based User Interaction Simulation on Real Infrastructure
John Ahlgren, Maria Eugenia Berezin, Kinga Bojarczuk, Elena Dulskyte, Inna Dvortsova, Johann George, Natalija Gucevska, Mark Harman, Ralf Lämmel, Erik Meijer, Silvia Sapora, Justin Spahr-Summers FACEBOOK Inc.
Abstract
We introduce the Web-Enabled Simulation (WES) research agenda, and describe FACEBOOK’s WW system. We describe the application of WW to reliability, integrity and privacy at FACEBOOK111‘FACEBOOK’, refers to the company, while ‘Facebook’ refers to the specific product., where it is used to simulate social media interactions on an infrastructure consisting of hundreds of millions of lines of code. The WES agenda draws on research from many areas of study, including Search Based Software Engineering, Machine Learning, Programming Languages, Multi Agent Systems, Graph Theory, Game AI, and AI Assisted Game Play. We conclude with a set of open problems and research challenges to motivate wider investigation.
中文速览
社交媒体平台每天承载数亿用户的复杂互动，仅靠传统软件测试很难发现那些由用户之间相互行为引发的"社会性漏洞"。为此，Facebook 提出了"网络使能仿真"（Web-Enabled Simulation，WES）框架，并基于此构建了名为 WW 的内部仿真系统——让大量自主 bot 在与真实平台完全相同的代码基础上模拟用户行为，同时与真实用户严格隔离。研究团队借助强化学习训练 bot 扮演"恶意行为者"，测试平台的可靠性、内容安全和隐私机制，并引入"社会测试"与"自动化机制设计"两种新方法，通过搜索最优交互机制来自动发现平台改进方案，而无需每次都修改底层代码。WW 已在数亿行代码规模的基础设施上验证了这一思路，论文同时梳理了一批开放性研究难题，旨在吸引机器学习、多智能体系统、搜索式软件工程等多个领域的研究者共同推进 WES 方法论的发展。
原文 arXiv:2004.05363;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2004.05363v1