Parallel Split-Join Networks for Shared Account Cross-Domain Sequential Recommendations

Ren, Pengjie; Lin, Yujie; Ma, Muyang; Chen, Zhumin; Zhang, Ren; Ma, Jun; Rijke, Maarten de

doi:10.1109/tkde.2021.3130927

Cited by 9 publications

(19 citation statements)

References 71 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the above works are mainly focused on users' static interactions, and user behaviors' sequential patterns and the shared-account characteristic are not considered. π-Net [3] and PSJNet [6] are two recently proposed methods for SCSR, but their studies are all based on RNNs, which are neither expressive enough to capture the multiple associations nor can model the structure information that bridges two domains.…”

Section: A Cross-domain Recommendationmentioning

confidence: 99%

“…Wen et al [36] address the shared-account challenge in a session-aware recommendation task, where a multi-user identification module drawing on the attention mechanism to distinguish behaviors of different users is proposed. The recently proposed methods π-Net [3] and PSJNet [6] argue that this task can be treated in an endto-end manner, and can also be improved by simultaneously considering the cross-domain information. In their approaches, each account is assumed to have H latent users, and the account representation is learned by merging all the userspecific representations of latent users within it.…”

Section: B Shared-account Recommendationmentioning

confidence: 99%

“…We evaluate our TiDA-GCN method on two real-world datasets that are released for the SCSR task, i.e., HVIDEO [3] and HAMAZON [6].…”

Section: B Datasetsmentioning

confidence: 99%

“…As the original Amazon data is not released for SCSR, the domain and account information cannot directly meet our requirement. Therefore, we follow the strategy in [6], and preprocess the Amazon data by the following steps: 1) As in our settings, users can simultaneously interact with items from different domains; we only retain the users who have interactions in both domains. 2) To simulate the shared-account characteristic, we first split the schedule into six intervals (i.e., 1996-2000, 2001-2003, 2004-2006, 2007-2009, 2010-2012, 2013-2015).…”

Section: B Datasetsmentioning

confidence: 99%

“…It solves a parallel sequential recommendation problem with a Gated Recurrent Unit (GRU)-based information-sharing network. Another related work is the PSJNet method [6]. It improves π-Net by using a split-join strategy on RNN.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Reinforcement Learning-enhanced Shared-account Cross-domain Sequential Recommendation

Guo

Zhang

Chen

et al. 2022

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

Shared-account Cross-domain Sequential Recommendation (SCSR) is an emerging yet challenging task that simultaneously considers the shared-account and cross-domain characteristics in the sequential recommendation. Existing works on SCSR are mainly based on Recurrent Neural Network (RNN) and Graph Neural Network (GNN) but they ignore the fact that although multiple users share a single account, it is mainly occupied by one user at a time. This observation motivates us to learn a more accurate user-specific account representation by attentively focusing on its recent behaviors. Furthermore, though existing works endow lower weights to irrelevant interactions, they may still dilute the domain information and impede the cross-domain recommendation. To address the above issues, we propose a reinforcement learning-based solution, namely RL-ISN, which consists of a basic cross-domain recommender and a reinforcement learning-based domain filter. Specifically, to model the account representation in the shared-account scenario, the basic recommender first clusters users' mixed behaviors as latent users, and then leverages an attention model over them to conduct user identification. To reduce the impact of irrelevant domain information, we formulate the domain filter as a hierarchical reinforcement learning task, where a high-level task is utilized to decide whether to revise the whole transferred sequence or not, and if it does, a low-level task is further performed to determine whether to remove each interaction within it or not. To evaluate the performance of our solution, we conduct extensive experiments on two real-world datasets, and the experimental results demonstrate the superiority of our RL-ISN method compared with the state-of-the-art recommendation methods.

show abstract

Section: A Cross-domain Recommendationmentioning

confidence: 99%

Section: B Shared-account Recommendationmentioning

confidence: 99%

“…We evaluate our TiDA-GCN method on two real-world datasets that are released for the SCSR task, i.e., HVIDEO [3] and HAMAZON [6].…”

Section: B Datasetsmentioning

confidence: 99%

Section: B Datasetsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%