A Survey on Sequential Recommendation (Frontiers of Computer Science, 2025)
This forum post discusses "A survey on sequential recommendation", published in November 2025 in *Frontiers of Computer Science* (2025).
Metadata
| Field | Content | |-------|---------| | Title | A survey on sequential recommendation | | Venue | Frontiers of Computer Science, 2025 | | Published | Nov 2025 | | Link | https://link.springer.com/article/10.1007/s11704-025-41329-w | | Type | Survey | | Section | Recommender Engine Surveys |
Background and Motivation
In large-scale search, recommendation, and personalization systems, recommendation has long faced challenges around efficiency, scalability, and understanding user intent. Traditional pipeline-style approaches tend to treat retrieval, ranking, and generation as separate stages, which makes it hard to satisfy the LLM-era combination of natural-language interaction, multi-hop reasoning, and real-time knowledge.
This survey is positioned at that intersection, aiming to systematically organize the theory and practice of sequential recommendation — modeling user behavior as ordered interaction sequences to predict future preferences.
Taxonomy of Methods
The post organizes the surrounding literature along several dimensions:
| Dimension | Sub-categories | Representative ideas | Strengths | Limitations | |-----------|----------------|----------------------|-----------|-------------| | Modeling paradigm | Discriminative / generative retrieval | Two-tower, cross-encoders, DSI, GPT indexing | Mature, scalable | Semantic drift, update cost | | LLM integration | RAG / Agent / Tool-use | Retrieval augmentation, search agents, API calls | Flexible, interpretable | Latency, error propagation | | Optimization objectives | Relevance / diversity / freshness | Multi-objective LTR, RLHF, online learning | Business-aligned | Scarce annotations | | Evaluation | Offline / online / human | nDCG, MRR, LLM-as-judge, A/B tests | Comparable | Gap vs. real satisfaction |
Evolution of the Field
- 2019–2021: BERT re-ranking and DPR established the foundations of neural retrieval.
- 2022–2023: RAG and FreshLLM drove the fusion of retrieval and generation.
- 2024 onward: Conversational/agentic search and generative recommender systems (Gen-RecSys) surged.
- 2025–2026: RL-trained search agents, Deep Research, and GraphRAG are emerging growth areas.
- Datasets: MS MARCO, BEIR, Natural Questions, domain corpora, public recommendation datasets
- Metrics: nDCG@10, MRR, Recall@k, Hit@k, human preference, task success rate, latency, token cost
- Baselines: BM25, dense retrieval, cross-encoder re-ranking, retrieval-free LLMs, commercial search APIs
- A Comprehensive Survey on Cross-Domain Recommendation
- A Comprehensive Survey on Retrieval Methods in Recommender Systems
- A Review of Modern Recommender Systems Using Generative Models (Gen-RecSys)
- A Survey of Generative Search and Recommendation in the Era of Large Language Models
- A Survey on LLM-powered Agents for Recommender Systems
- Original paper: *A survey on sequential recommendation*, Frontiers of Computer Science, 2025. https://link.springer.com/article/10.1007/s11704-025-41329-w
On the recommendation side specifically, the field has progressed from matrix factorization, deep CTR models, and sequential Transformers, toward LLM instruction-following and generative recommendation — with persistent challenges of sparse user behavior, huge item catalogs, and multi-objective business trade-offs.
Evaluation Practices
Typical benchmarks and metrics covered in this line of research include:
Key Insights
1. Architecture: Cascaded retrieval + re-ranking + generation remains mainstream, but agentic paradigms make the retrieval policy itself learnable. 2. Data: High-quality instruction data and click/session logs are both critical; synthetic data risks leakage and distribution shift. 3. Evaluation: The gap between offline metrics and online satisfaction is widening; LLM-as-judge needs cross-validation with human evaluation. 4. Productization: Latency, cost, interpretability, and safety are hard constraints for industrial deployment.
Open Problems
The post highlights: lack of unified benchmarks, non-reproducible private data, LLM evaluation bias, and the safety and cost constraints of agentic systems. Future directions include finer-grained process supervision, joint retrieval–reasoning training, enterprise metadata governance, and multimodal and cross-lingual consistency.