← 索引

研究索引

GEO / AEO / AI Search 领域的公开研究(2023–2026)。

分两部分:我们按问题分组的精选(带着判断读),和完整清单(按主题编,可检索)。


一、按八个问题分组(精选)

1 · 这件事值不值得做

研究它说明了什么出处
How Generative AI Disrupts Search: Consumer Behavior, Website Traffic, and the Impact of Google AI Overviews用真实日志数据,量化 AI 摘要对用户行为和网站流量的影响ACM SIGIR 2026
Zero-Click Search and Answer Engines: A Literature Review系统性梳理"零点击"趋势下搜索可见性的转型JIEM, 2025
Characterizing Web Search in the Age of Generative AI对比传统搜索与生成式搜索的检索足迹与稳定性差异arXiv:2510.11560
GEO: Generative Engine Optimization这个领域的奠基论文,提出 GEO 的问题、方法与评测基准KDD 2024 / arXiv:2311.09735

2 · "对题"为什么比"权威"更重要

研究它说明了什么出处
From Citation Selection to Citation Absorption提出"引用选择 → 引用吸收"两阶段框架;与引用影响力相关性最高的是内容与问题的匹配度,高于内容质量本身arXiv:2604.25707
C-SEO Bench: Does Conversational SEO Work?建立基准,系统检验"对话式优化"到底有没有效果NeurIPS Datasets & Benchmarks 2025 / arXiv:2506.11097
AI Answer Engine Citation Behavior: A Longitudinal Study of GEO in 16 Generative Engines16 个生成式引擎的引用行为纵向研究arXiv:2509.10762
GEO at Scale: Measuring Brand Visibility Across AI Search Engines跨引擎的品牌可见度规模化测量arXiv:2606.20065

3 · "引用广度"和"吸收深度"是两件事

研究它说明了什么出处
From Citation Selection to Citation Absorption引得多不等于影响大:有的引擎引用条数多但影响力弱,有的引用条数少但影响强——两个指标必须分开测量arXiv:2604.25707
AI Answer Engine Citation Behavior同一内容在不同引擎里的引用行为不一样,单平台结论不能外推arXiv:2509.10762
When Content is Goliath and Algorithm is David: The Style and Semantic Effects of Generative Search Engine内容风格与语义结构,各自如何影响生成式搜索的取舍ResearchGate, 2025

4 · 内容写成什么样,才会被引用

研究它说明了什么出处
What Generative Search Engines Like: Explaining and Optimizing Website Visibility解释生成式搜索偏好什么样的页面,并给出可操作的优化方向arXiv:2510.11438
Structural Feature Engineering for Generative Engine Optimization页面结构特征与引用效果之间的关系——结构本身有独立贡献arXiv:2603.29979
Think Before Writing: Feature-Level Optimization for Generative Engine Optimization在写作之前做特征级优化,比写完再改有效arXiv:2604.19113
What Evidence Do Language Models Find Convincing?模型认为什么样的证据可信——直接影响内容该怎么写arXiv:2402.11782

5 · 怎么测才可信(方法问题)

研究它说明了什么出处
Don't Measure Once: Reliable AI Search Visibility Requires Repeated Trials and Structure-Aware Metrics单次测量不可靠——可见性评估需要重复试验与结构感知指标arXiv:2604.07585
Disentangling Answer Engine Optimization from Platform Growth: A Log-Based Natural Experiment on ChatGPT Referral Traffic用日志自然实验,把"优化带来的变化"与"平台自身的增长"分开arXiv:2606.04362
SAGEO: A Search Arena for Generative Engine Optimization in a Realistic Environment更接近真实环境的评测场,用于对比不同做法arXiv:2602.12187
SAGE: Scalable AI Governance & Evaluation生产级搜索的评估与治理框架arXiv:2602.07840

6 · "引到错的版本"是已被研究过的风险

研究它说明了什么出处
CONFLICTBANK: A Benchmark for Evaluating the Influence of Knowledge Conflicts in LLMs当模型同时检索到新旧两版信息时,回答会不稳定——知识冲突可以被测量OpenReview, 2024
Persistent Pre-Training Poisoning of LLMs预训练阶段的污染会长期留存,且难以事后清除arXiv:2410.13722
Adversarial Search Engine Optimization for Large Language Models对抗式优化(把内容"塞给"模型)的手法与机理arXiv:2406.18382
Manipulating Large Language Models to Increase Product Visibility操纵内容以提升可见度的具体路径arXiv:2404.07981

7 · 合规做法是可行的

研究它说明了什么出处
White Hat Search Engine Optimization using Large Language Models用合规手段做内容优化,效果有实证依据arXiv:2502.07315
Unveiling the Resilience of LLM-Enhanced Search Engines Against Adversarial Attacks模型侧对操纵手段的抗性正在增强——捷径的窗口在收窄arXiv:2603.25500
风险与操纵主题(共 10 篇)刷量、投毒、排名操纵的手法与检测研究见下方完整清单 06 类

8 · 要做底座,技术上怎么搭

研究它说明了什么出处
MCompassRAG: Topic Metadata as a Semantic Compass for Paragraph-Level Retrieval主题元数据能显著改善段落级检索——"分类"这件事在技术上是有效的arXiv:2606.18508
Decoupling Search from Reasoning: A Vendor-Agnostic Grounding Architecture for LLM Agents把"检索"与"推理"解耦,做与厂商无关的可信来源层arXiv:2606.18947
SproutRAG: Attention-Guided Tree Search for Long-Document RAG长文档场景下的检索优化arXiv:2606.18381

二、完整清单(54 篇 · 按 10 个主题)

01 · GEO 基础框架(4)

标题年来源
GEO: Generative Engine Optimization2023/24arXiv:2311.09735 · KDD 2024
Generative Engine Optimization in Digital Repositories2025Infonomy
Generative Engine Optimization: How to Dominate AI Search2025arXiv:2509.08919
Navigating the Shift: Web Search vs Generative AI Response Generation2026arXiv:2601.16858

02 · GEO 方法优化(7)

标题年来源
Beyond Keywords: GSEO with Content-Centric Agents2025arXiv:2509.05607
Beyond SEO: A Transformer-Based Approach for Web Content Optimisation2025arXiv:2507.03169
IF-GEO: Conflict-Aware Instruction Fusion for Multi-Query GEO2026arXiv:2601.13938
Role-Augmented Intent-Driven GSEO2025arXiv:2508.11158
Think Before Writing: Feature-Level Optimization for GEO2026arXiv:2604.19113
What Generative Search Engines Like2025arXiv:2510.11438
White Hat Search Engine Optimization using LLMs2025arXiv:2502.07315

03 · GEO 测量评估(8)

标题年来源
AI Answer Engine Citation Behavior: A Longitudinal Study in 16 Generative Engines2025arXiv:2509.10762
C-SEO Bench: Does Conversational SEO Work?2025arXiv:2506.11097 · NeurIPS D&B
Don't Measure Once: Reliable AI Search Visibility Requires Repeated Trials2026arXiv:2604.07585
Disentangling AEO from Platform Growth: Log-Based Natural Experiment on ChatGPT Referral2026arXiv:2606.04362
From Citation Selection to Citation Absorption2026arXiv:2604.25707
GEO at Scale: Measuring Brand Visibility Across AI Search Engines2026arXiv:2606.20065
SAGEO: A Search Arena for GEO in a Realistic Environment2026arXiv:2602.12187
Structural Feature Engineering for GEO2026arXiv:2603.29979

04 · AI 搜索实证(6)

标题年来源
A Survey of Generative Search and Recommendation in the Era of LLMs2024/25generative-rec.github.io
Characterizing Web Search in the Age of Generative AI2026arXiv:2510.11560
What Do Chinese-Language Generative Search Engines Cite and Surface?2026arXiv:2607.15771
How Generative AI Disrupts Search: Consumer Behavior, Website Traffic, Google AI Overviews2026ACM SIGIR 2026
NExT-Search: Rebuilding User Feedback Ecosystem for Generative AI Search2025arXiv:2505.14680
What Evidence Do Language Models Find Convincing?2024arXiv:2402.11782

05 · AEO 理论整合(5)

标题年来源
From SEO to AEO: Generative AI and the Transformation of Search Visibility2025ResearchGate
Optimizing for the AI-Driven Search Era: Integrated Framework for SEO/GEO/AEO2025IJSREM
Smart Search Optimization: Integrating SEO, AEO and GEO2025Loro Journal
How the Transition from SEO to GEO, AEO, and AIO Impacts Digital Marketing2025IJCRT
Zero-Click Search and Answer Engines: A Literature Review2025/26JIEM

06 · 风险、操纵与对抗(10)

标题年来源
Adversarial Search Engine Optimization for LLMs2024arXiv:2406.18382
CONFLICTBANK: Knowledge Conflicts in LLMs2024OpenReview
Dynamics of Adversarial Attacks on LLM-Based Search Engines2025arXiv:2501.00745
GASLITEing the Retrieval: Vulnerabilities in Dense Embedding-Based Search2024/25OpenReview
Manipulating LLMs to Increase Product Visibility2024arXiv:2404.07981
Persistent Pre-Training Poisoning of LLMs2024arXiv:2410.13722
PoisonArena: Competing Poisoning Attacks in RAG2025arXiv:2505.12574
Ranking Manipulation for Conversational Search Engines2024arXiv:2406.03589
StealthRank: LLM Ranking Manipulation via Stealthy Prompt Optimization2025arXiv:2504.05804
Unveiling the Resilience of LLM-Enhanced Search Engines Against Adversarial Attacks2026arXiv:2603.25500

07 · 垂直场景与多模态(5)

标题年来源
Caption Injection for Optimization in Generative Search Engines2025arXiv:2511.04080
E-GEO: A Testbed for GEO in E-Commerce2025arXiv:2511.20867
Multimodal GEO: Rank Manipulation for VLM Rankers2026ResearchGate
GEO: A VLM and Agent Framework for Pinterest Acquisition Growth2026arXiv:2602.02961
When Content is Goliath and Algorithm is David: Style and Semantic Effects2025ResearchGate

08 · AI 搜索架构与 Agentic Search(4)

标题年来源
Decoupling Search from Reasoning: A Vendor-Agnostic Grounding Architecture2026arXiv:2606.18947
Beyond Parallel Sampling: Diverse Query Initialization for Agentic Search2026arXiv:2606.17209
ScholarQuest: Taxonomy-Guided Benchmark for Agentic Academic Paper Search2026arXiv:2606.20235
When Does Streaming Tool Use Help? Tool-Intent Stabilization in Streaming RAG2026arXiv:2606.20113

09 · RAG 检索优化(3)

标题年来源
MCompassRAG: Topic Metadata as a Semantic Compass for Paragraph-Level Retrieval2026arXiv:2606.18508
SCAR: Semantic Continuity-Aware Retrieval for RAG2026arXiv:2606.16661
SproutRAG: Attention-Guided Tree Search for Long-Document RAG2026arXiv:2606.18381

10 · 搜索评估与治理(2)

标题年来源
Scaling Search Relevance: Augmenting App Store Ranking with LLM-Generated Judgments2026arXiv:2602.23234
SAGE: Scalable AI Governance & Evaluation2026arXiv:2602.07840

三、怎么用这份索引

  1. 按问题读——先看第一部分,八个问题对应八组文献;
  2. 按主题查——第二部分按 10 个主题编,可直接用 arXiv 编号取原文;
  3. 顺手核——所有出处都是公开可复查的;引用时注明来源即可。

各篇著作权归原作者或出版方,本索引只做编目与指向,不转载原文。