Organizations: School of Big Data and Software Engineering, Chongqing University, Chongqing, 401331, China · College of Computer Science and Engineering, Chongqing University of Technology, Chongqing, 401135, China · School of Computing and Mathematical Sciences, University of Leicester, Leicester, 401135, United Kingdom · Department of Plastic Surgery, Chongqing University Central Hospital, Chongqing Emergency Medical Center, Chongqing, 400000, China · College of Environmental Science and Engineering, Tongji University, Shanghai, 200092, China
Shared-account usage is common on streaming and e-commerce platforms, where multiple users share one account. Existing shared-account sequential recommendation (SSR) methods often assume a fixed number of latent users per account, limiting their ability to adapt to diverse sharing patterns and reducing recommendation accuracy. Recent latent reasoning technique applied in sequential recommendation (SR) generate intermediate embeddings from the user embedding (e.g, last item embedding) to uncover users' potential interests, which inspires us to treat the problem of inferring the number of latent users as generating a series of intermediate embeddings, shifting from inferring preferences behind user to inferring the users behind account. However, the last item cannot be directly used for reasoning in SSR, as it can only represent the behavior of the most recent latent user, rather than the collective behavior of the entire account. To address this, we propose DisenReason, a two-stage reasoning method tailored to SSR. DisenReason combines behavior disentanglement stage from frequency-domain perspective to create a collective and unified account behavior representation, which serves as a pivot for latent user reasoning stage to infer the number of users behind the account. Experiments on four benchmark datasets show that DisenReason consistently outperforms all state-of-the-art baselines across four benchmark datasets, achieving relative improvements of up to 12.56% in MRR@5 and 6.06% in Recall@20.
Figures & tables
Figure 1: Visualization of frequency-domain disentanglement for shared-account behavior sequence.
Notation
Definition
A
Set of shared accounts
Aj
The j -th shared account
V
Set of items
vi
The i -th item
M
Interaction matrix where Mij=1 indicates account Ai interacted with item vj
Sk=[v1k,v2k,…,vsk]
Interaction sequence of account Ak in chronological order
Table 1: Notations used in DisenReason
Figure 2: Overview of the proposed DisenReason framework. This figure shows the details of DisenReason: stage one focuses on disentangling the mixed behavioral sequence to create a unified account pivot, while stage two progressively infers latent users based on this pivot.
HVIDEO
HAMAZON
E-domain
V-domain
M-domain
B-domain
Items
8,367
11,404
67,161
126,547
Interactions
2,129,500
1,893,784
2,196,574
2,135,995
Avg. sequence length
15.85
14.09
15.27
14.84
Accounts
13,714
13,724
Sequences
134,349
143,885
Table 2: Statistical details for HVIDEO and HAMAZON datasets.
Models
HV-E
HV-V
HA-M
HA-B
RC@5
MRR@5
RC@5
MRR@5
RC@5
MRR@5
RC@5
MRR@5
w/o LightGCN
61.96
51.32
79.84
67.68
48.61
48.52
49.44
49.35
w/o Behavior Disentanglement
55.09
33.35
77.81
53.17
35.89
22.62
36.21
22.67
w/o Adaptive Fusion
62.51
50.88
81.59
67.64
49.05
48.76
49.54
49.39
w/o Residual Operation
63.32
51.37
81.34
67.69
49.08
48.84
49.50
49.32
Our
63.43
52.05
81.89
67.89
49.13
48.81
49.56
49.39
Table 4: Ablation study of DisenReason on four datasets. ‘w/o’ denotes the model variant without the corresponding design.
Figure 3: Hyperparameter analysis of DisenReason on HV-V and HV-E dataset.
Figure 4: Performance comparison of DisenReason and baselines under different sequence lengths and training data scales on HV-V and HV-E datasets.
Figure 5: Latent user number per account ID on HV-E and HV-V dataset. It illustrates the number of latent users inferred for each account ID through the reasoning process.
This paper addresses two persistent challenges in sequential recommendation: (i) evidence insufficiency-cold-start sparsity together with noisy, length-varying item texts; and (ii) opaque modeling of dynamic, multi-faceted intents across long/short horizons. We propose R3-REC (Reasoning-Retrieval-Recommendation), a prompt-centric, retrieval-augmented framework that unifies Multi-level User Intent Reasoning, Item Semantic Extraction, Long-Short Interest Polarity Mining, Similar User Collaborative Enhancement, and Reasoning-based Interest Matching and Scoring. Across ML-1M, Games, and Bundle, R3-REC consistently surpasses strong neural and LLM baselines, yielding improvements up to +10.2% (HR@1) and +6.4% (HR@5) with manageable end-to-end latency. Ablations corroborate complementary gains of all modules.
Yuchen Miao, Mingxuan Cui, Yitong Zhu +2
Sydney Smart Technology College, Northeastern University, China
Sequential recommender systems typically infer user preferences through single-pass encoding of interaction histories without iterative refinement, relying on increasingly deep architectures to capture complex patterns. In this work, we revisit sequential recommendation from a recursive inference perspective: can user preferences be modeled as a persistent latent state that is recursively refined? We propose RecRec (Recursive Recommendation), a lightweight model that maintains a compact latent state and updates it through a shared recursive module conditioned on interaction evidence. Unlike prior recursive models, RecRec introduces an evidence-anchored correction mechanism that stabilizes refinement by grounding each update in the original interaction context, preventing semantic drift during deep recursive reasoning. Experiments on three benchmark datasets under standard evaluation protocols show that RecRec matches or outperforms state-of-the-art sequential, graph-based, and reasoning-enhanced recommenders while using only 3.9M to 14M parameters. Ablation studies demonstrate that both recursive refinement and the evidence-anchored correction gate contribute significantly to performance, highlighting the effectiveness of recursive latent inference as a scalable alternative to deeper or language-based architectures. Code is available at https://anonymous.4open.science/r/RecRec-6B67/README.md.
Large Language Models (LLMs) have shown strong potential for recommendation by leveraging their semantic understanding and contextual modeling capabilities. Recent studies further introduce reasoning mechanisms to improve user preference modeling. However, explicit natural-language reasoning incurs substantial inference overhead, whereas existing latent reasoning methods mainly focus on generating or verifying intermediate states, leaving their layer-wise preference roles and contributions insufficiently characterized. We propose HiLaR, a Hierarchical Latent Reasoning framework with layer-aware reinforcement optimization for LLM-based recommendation. HiLaR constructs temporal-guided hierarchical user preference representations, aligns them with multiple LLM latent reasoning states, and organizes the reasoning process from broad preferences to fine-grained current intents. To further optimize the reasoning trajectory, HiLaR combines final recommendation feedback with layer-aware process rewards derived from the marginal target-likelihood gain of each state. Experiments on four Amazon benchmark datasets show that HiLaR generally outperforms strong sequential, generative, and LLM-based recommendation baselines. Ablation and sensitivity analyses further verify the contribution of hierarchical representation learning, latent alignment, and process-level optimization. Our code is available in https://github.com/hupeiyu21/HiLaR.
Peiyu Hu, Siying Gu, Weihai Lu +8
1Xi’an Jiaotong-Liverpool University · 2Xiaohongshu · 3Peking University +1