Search-o1: Agentic Search-Enhanced Large Reasoning Models (EMNLP 2025)
Metadata
| Field | Value | |---|---| | Title | Search-o1: Agentic Search-Enhanced Large Reasoning Models | | Venue | EMNLP 2025 Main Conference | | Link | https://aclanthology.org/2025.emnlp-main.276 | | Type | Academic paper | | Category | Agentic Search |
One-line summary
Search-o1 is an academic work on agentic search-enhanced large reasoning models, targeting the intersection of large-scale search, recommendation, and personalization systems in the LLM era.
Background and Motivation
Agentic search in large-scale search, recommendation, and personalization systems has long faced challenges in efficiency, scalability, and user-intent understanding. Traditional pipeline approaches separate retrieval, ranking, and generation, making it hard to meet modern demands for natural-language interaction, multi-hop reasoning, and real-time knowledge. Search-o1 is proposed in this context to advance the theoretical and practical boundaries of this intersection.
The problem scope includes:
- Open-domain information access
- Enterprise knowledge retrieval
- Conversational search
- Semantic understanding in recommendation systems
- End-to-end architectures that coordinate external knowledge sources with generative models
- A unified perspective that organizes scattered related work into a comparable framework.
- A clear decomposition of method components: representation learning, retrievers, rerankers, planners, generators, and feedback mechanisms, easing engineering adoption.
- Reproducible benchmarks, datasets, or taxonomy tables, lowering the entry cost for follow-up research.
- Discussion of interfaces with emerging paradigms such as LLM tool calling, reinforcement learning, and multi-agent collaboration, and migration paths from research prototypes to industrial systems.
- Explicitly listed open problems: evaluation trustworthiness, latency and cost, hallucination and safety, and cross-lingual / multimodal extension.
- Datasets: MS MARCO, BEIR, Natural Questions, domain-specific corpora, public recommendation sets
- Metrics: nDCG@10, MRR, Recall@k, Hit@k, human preference, task success rate, latency, and token cost
- Baselines: BM25, dense retrieval, cross-encoder reranking, retrieval-free LLMs, commercial search APIs
- Ablations: contribution of retrieval steps, reranking depth, and training data scale
- A Systematic Framework for Enterprise Knowledge Retrieval
- Retrieval Augmented Generation and Understanding in Vision: A Survey
- Synergizing RAG and Reasoning: A Systematic Review (arXiv:2504.15909)
- AceSearcher: Bootstrapping Reasoning and Search for LLMs via Reinforcement Learning (arXiv:2509.24193)
- Agentic Information Retrieval (arXiv:2410.09713)
- Original paper: Search-o1: Agentic Search-Enhanced Large Reasoning Models. https://aclanthology.org/2025.emnlp-main.276
Core Contributions
Method / System Architecture
The methodological pattern typically follows four steps: problem formalization → model/system design → training/construction pipeline → inference pipeline.
1. Input and representation: encode queries, documents, and user context as dense/sparse representations or structured prompts. 2. Core modules: retrievers, rerankers, planners, memory modules, and tool interfaces, chained or combined per task. 3. Learning strategies: supervised fine-tuning, contrastive learning, distillation, reinforcement learning (including process rewards), and bootstrapped data synthesis. 4. Inference strategies: single-turn retrieval, iterative retrieval, parallel sub-queries, early stopping, and budget control.
Experiments and Evaluation
Typical experimental design covered in this line of work:
> Note: The original post states that specific numerical results should be verified against the original PDF; quantitative claims in this summary follow the source's experimental-design framing.
Key Takeaways for Search / Rec / Personalization
1. Architecture: cascaded retrieval + rerank + generation remains mainstream, but the agentic paradigm makes retrieval frequency and policy themselves learnable. 2. Data: high-quality instruction data and click/session logs are both critical; synthetic data requires protection against knowledge leakage and distribution shift. 3. Evaluation: the gap between offline metrics and online satisfaction is widening; LLM-as-judge needs cross-validation against human evaluation. 4. Product: latency, cost, explainability, and safety policies are hard constraints for industrial deployment, not just academic benchmarks.
Limitations and Future Work
Possible limitations include experiment scale bounded by GPU budget, mismatch between benchmarks and real user distributions, unknown cross-lingual generalization from English-centric data, and safety risks of agents on the open web. Future directions: more efficient test-time compute allocation, deeper integration with knowledge graphs and structured databases, and causal/fairness constraints for recommendation systems.
Related Entries
Engineering Deployment Checklist
| Check | Question | Suggestion | |---|---|---| | Data | Does training/index contain PII? How is versioning handled? | Partitioned indexes, anonymization, rollback-capable embedding versions | | Latency | What is the p99 budget? How many retrieval steps? | Cascades + early stopping, caching hot queries, async reranking | | Quality | Do offline gains translate to online CTR/satisfaction? | Interleaving experiments, human audit samples, citation verification | | Safety | Does open retrieval introduce poisoning/bias? | Source whitelists, adversarial detection, output filtering | | Cost | Token and GPU cost per query? | Route to smaller models, distillation, hybrid sparse+dense |
Glossary
| Term | Meaning | |---|---| | IR | Information Retrieval | | RAG | Retrieval-Augmented Generation | | LTR | Learning to Rank | | nDCG | Normalized Discounted Cumulative Gain, a ranking-quality metric | | Agentic Search | Modeling search as sequential decision-making and tool calling by an agent | | Gen-IR | Generative Information Retrieval |