2026-03-07 周六 · Agentic

独立报告页 · 在站内浏览（含往期切换）→
Generative生成式推荐 · 10 篇Personalization个性化推荐 · 4 篇AI SearchAI 搜索 · 10 篇Agentic智能体搜索 · 10 篇Infra系统与索引 · 10 篇

今日场景社区搜索 (UGC 短文本搜索): 用户在社区平台搜索帖子、问答、评论等 UGC 内容。核心指标是搜索相关性 (nDCG) 和点击率 (CTR)。关注：语义匹配、query 理解、短文本召回、排序模型、LLM 在搜索中的应用。 — 评分 10 篇，入选 10 篇，分数 3.00 – 3.00

本期目录 · 10 篇
  1. 01SearchGym: A Modular Infrastructure for Cross-Platform Benchmarking and Hybrid Search Orchestration
  2. 02FinRetrieval: A Benchmark for Financial Data Retrieval by AI Agents
  3. 03Signal in the Noise: Decoding the Reality of Airline Service Quality with Large Language Models
  4. 04Still Fresh? Evaluating Temporal Drift in Retrieval Benchmarks
  5. 05DARE: Aligning LLM Agents with the R Statistical Ecosystem via Distribution-Aware Retrieval
  6. 06Scaling Laws for Reranking in Information Retrieval
  7. 07Beyond Text: Aligning Vision and Language for Multimodal E-Commerce Retrieval
  8. 08Detecting RAG Advertisements Across Advertising Styles
  9. 09Debiasing Sequential Recommendation with Time-aware Inverse Propensity Scoring
  10. 10Core-based Hierarchies for Efficient GraphRAG
LEAD STORY · 今日必读

SearchGym: A Modular Infrastructure for Cross-Platform Benchmarking and Hybrid Search Orchestration

arXiv:2603.04402v1 Announce Type: new Abstract: The rapid growth of Retrieval-Augmented Generation (RAG) has created a proliferation of toolkits, yet a fundamental gap remains between experimental pr

arXiv ↗
相关 3新颖 3实用 3证据 3时效 3
3.00
综合分 / 5
偏好调整 3.00
KEY INSIGHT · 对业务意味着什么N/A - auto-generated placeholder (LLM scoring unavailable)业务关联 N/A - auto-generated placeholder证据 N/A - auto-generated placeholder
Jerome Tze-Hou Hsu · Fri, 06 Ma · arXiv

今日其余入选

9 篇 · 按综合排名
02 3.00

FinRetrieval: A Benchmark for Financial Data Retrieval by AI Agents

arXiv:2603.04403v1 Announce Type: new Abstract: AI agents increasingly assist with financial research, yet no benchmark evaluates their ability to retrieve specific numeric values from structured dat

INSIGHTN/A - auto-generated placeholder (LLM scoring unavailable)

业务关联N/A - auto-generated placeholder

证据N/A - auto-generated placeholder

Eric Y. Kim, Jie Huang
arXiv·Fri, 06 Ma·Relevance 3 · Novelty 3 · Practical 3 · Evidence 3 · Timely 3
03 3.00

Signal in the Noise: Decoding the Reality of Airline Service Quality with Large Language Models

arXiv:2603.04404v1 Announce Type: new Abstract: Traditional service quality metrics often fail to capture the nuanced drivers of passenger satisfaction hidden within unstructured online feedback. Thi

INSIGHTN/A - auto-generated placeholder (LLM scoring unavailable)

业务关联N/A - auto-generated placeholder

证据N/A - auto-generated placeholder

Ahmed Dawoud, Osama El-Shamy, Ahmed Habashy
arXiv·Fri, 06 Ma·Relevance 3 · Novelty 3 · Practical 3 · Evidence 3 · Timely 3
04 3.00

Still Fresh? Evaluating Temporal Drift in Retrieval Benchmarks

arXiv:2603.04532v1 Announce Type: new Abstract: Information retrieval (IR) benchmarks typically follow the Cranfield paradigm, relying on static and predefined corpora. However, temporal changes in t

INSIGHTN/A - auto-generated placeholder (LLM scoring unavailable)

业务关联N/A - auto-generated placeholder

证据N/A - auto-generated placeholder

Nathan Kuissi, Suraj Subrahmanyan, Nandan Thakur, Jimmy Lin
arXiv·Fri, 06 Ma·Relevance 3 · Novelty 3 · Practical 3 · Evidence 3 · Timely 3
05 3.00

DARE: Aligning LLM Agents with the R Statistical Ecosystem via Distribution-Aware Retrieval

arXiv:2603.04743v1 Announce Type: new Abstract: Large Language Model (LLM) agents can automate data-science workflows, but many rigorous statistical methods implemented in R remain underused because

INSIGHTN/A - auto-generated placeholder (LLM scoring unavailable)

业务关联N/A - auto-generated placeholder

证据N/A - auto-generated placeholder

Maojun Sun, Yue Wu, Yifei Xie, Ruijian Han, Binyan Jiang, Defeng Sun, Yancheng Yuan, Jian Huang
arXiv·Fri, 06 Ma·Relevance 3 · Novelty 3 · Practical 3 · Evidence 3 · Timely 3
06 3.00

Scaling Laws for Reranking in Information Retrieval

arXiv:2603.04816v1 Announce Type: new Abstract: Scaling laws have been observed across a wide range of tasks, such as natural language generation and dense retrieval, where performance follows predic

INSIGHTN/A - auto-generated placeholder (LLM scoring unavailable)

业务关联N/A - auto-generated placeholder

证据N/A - auto-generated placeholder

Rahul Seetharaman, Aman Bansal, Hamed Zamani, Kaustubh Dhole
arXiv·Fri, 06 Ma·Relevance 3 · Novelty 3 · Practical 3 · Evidence 3 · Timely 3
07 3.00

Beyond Text: Aligning Vision and Language for Multimodal E-Commerce Retrieval

arXiv:2603.04836v1 Announce Type: new Abstract: Modern e-commerce search is inherently multimodal: customers make purchase decisions by jointly considering product text and visual informations. Howev

INSIGHTN/A - auto-generated placeholder (LLM scoring unavailable)

业务关联N/A - auto-generated placeholder

证据N/A - auto-generated placeholder

Qujiaheng Zhang, Guagnyue Xu, Fengjie Li
arXiv·Fri, 06 Ma·Relevance 3 · Novelty 3 · Practical 3 · Evidence 3 · Timely 3
08 3.00

Detecting RAG Advertisements Across Advertising Styles

arXiv:2603.04925v1 Announce Type: new Abstract: Large language models (LLMs) enable a new form of advertising for retrieval-augmented generation (RAG) systems in which organic responses are blended w

INSIGHTN/A - auto-generated placeholder (LLM scoring unavailable)

业务关联N/A - auto-generated placeholder

证据N/A - auto-generated placeholder

Sebastian Heineking, Wilhelm Pertsch, Ines Zelch, Janek Bevendorff, Benno Stein, Matthias Hagen, Mar...
arXiv·Fri, 06 Ma·Relevance 3 · Novelty 3 · Practical 3 · Evidence 3 · Timely 3
09 3.00

Debiasing Sequential Recommendation with Time-aware Inverse Propensity Scoring

arXiv:2603.04986v1 Announce Type: new Abstract: Sequential Recommendation (SR) predicts users next interactions by modeling the temporal order of their historical behaviors. Existing approaches, incl

INSIGHTN/A - auto-generated placeholder (LLM scoring unavailable)

业务关联N/A - auto-generated placeholder

证据N/A - auto-generated placeholder

Sirui Huang, Jing Long, Qian Li, Guandong Xu, Qing Li
arXiv·Fri, 06 Ma·Relevance 3 · Novelty 3 · Practical 3 · Evidence 3 · Timely 3
10 3.00

Core-based Hierarchies for Efficient GraphRAG

arXiv:2603.05207v1 Announce Type: new Abstract: Retrieval-Augmented Generation (RAG) enhances large language models by incorporating external knowledge. However, existing vector-based methods often f

INSIGHTN/A - auto-generated placeholder (LLM scoring unavailable)

业务关联N/A - auto-generated placeholder

证据N/A - auto-generated placeholder

Jakir Hossain, Ahmet Erdem Sar{\i}y\"uce
arXiv·Fri, 06 Ma·Relevance 3 · Novelty 3 · Practical 3 · Evidence 3 · Timely 3

其他赛道今日

点击切换
Generative
10
生成式推荐 · 篇入选Personalization
4
个性化推荐 · 篇入选AI Search
10
AI 搜索 · 篇入选Infra
10
系统与索引 · 篇入选