Deep Session Interest Network for Click-Through Rate Prediction

Feng, Yufei; Lv, Fuyu; Shen, Weichen; Wang, Menghan; Sun, Fei; Zhu, Yu; Yang, Keping

doi:10.48550/arxiv.1905.06482

Cited by 51 publications

(55 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Deep Interest Evolution Network (DIEN) [35] assumes that user interests is dynamic, and thus capture evoloving user interest from their historical behaviors on items via a GRU network with attentional update gates. Deep Session Interest Network (DSIN) [7] observes that user behaviors can be grouped by different sessions, so it leverages Bi-LSTM with self-attention layers to model the inter-session and intro-session interests of users. However, although these models try to use powerful network architectures to model different kinds of historical behaviors, they did not make user of multi-source neighbourhood information, which limits their effectiveness.…”

Section: Related Work 51 Ctr Predictionmentioning

confidence: 99%

“…However, this learning paradigm treats the sparse categorical feature equally and ignores the intrinsic structures among them, e.g., the sequential order of historical behaviors. Recently, several studies in user interests modeling [7,18,19,35,36] emphasize on the sequential structure of user behaviour features. They model the historical items of users as sequences and exploit the sequence modeling methods such as LSTM [11], GRU [3] and multi-head attention [25] to effectively model the user preference.…”

mentioning

confidence: 99%

“…They model the historical items of users as sequences and exploit the sequence modeling methods such as LSTM [11], GRU [3] and multi-head attention [25] to effectively model the user preference. Typical methods include DIN [36], DIEN [35], DSIN [7], SDM [18] and DMR [19], etc. Although existing methods for CTR prediction have achieved significant progress, the above methods only focus on mining the interaction between the candidate item and the user's historical behaviours, which suffers from two limitations: On the one hand, user behaviours might be sparse for inactive users, which rise a cold-start problem and impede the quality to representation.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Neighbour Interaction based Click-Through Rate Prediction via Graph-masked Transformer

Min¹,

Yu²,

Xu³

et al. 2022

Preprint

View full text Add to dashboard Cite

Click-Through Rate (CTR) prediction, is an essential component of online advertising. The mainstream techniques mostly focus on feature interaction or user interest modeling, which rely on users' directly interacted items. The performance of these methods is usually impeded by inactive behaviours and system's exposure, incurring that the features extracted do not contain enough information to represent all potential interests. For this sake, we propose Neighbor-Interaction based CTR prediction, which put this task into a Heterogeneous Information Network (HIN) setting, then involves local neighborhood of the target user-item pair in the HIN to predict their linkage. In order to enhance the representation of the local neighbourhood, we consider four types of topological interaction among the nodes, and propose a novel Graph-masked Transformer architecture to effectively incorporates both feature and topological information. We conduct comprehensive experiments on two real world datasets and the experimental results show that our proposed method outperforms state-of-the-art CTR models significantly. CCS CONCEPTS• Information systems → Computational advertising.

show abstract

Section: Related Work 51 Ctr Predictionmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Neighbour Interaction based Click-Through Rate Prediction via Graph-masked Transformer

Min¹,

Yu²,

Xu³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Single behavior modeling methods mainly include RNN series (such as DIEN [14], LSTM and context information [15], etc. ), CNN [16] and Attention mechanism (such as DIN [17], DSIN [18], etc.). Multi behavior modeling methods include collective matrix factorization (CMF) [19,20] and modeling into deep semantic spaces together (such as ATRank [11], NMTR [9], CSAN [10], EHCF [31], MBGCN [32], etc.).…”

Section: Behavior Modelingmentioning

confidence: 99%

AskMe: Joint Individual-level and Community-level Behavior Interaction for Question Recommendation

Guo

Liu

et al. 2021

Preprint

View full text Add to dashboard Cite

Questions in Community Question Answering (CQA) sites are recommended to users, mainly based on users' interest extracted from questions that users have answered or have asked. However, there is a general phenomenon that users answer fewer questions while pay more attention to follow questions and vote answers. This can impact the performance when recommending questions to users (for obtaining their answers) by using their historical answering behaviors on existing studies. To address the data sparsity issue, we propose AskMe, which aims to leverage the rich, hybrid behavior interactions in CQA to improve the question recommendation performance. On the one hand, we model the rich correlations between the user's diverse behaviors (e.g., answer, follow, vote) to obtain the individual-level behavior interaction. On the other hand, we model the sophisticated behavioral associations between similar users to obtain the community-level behavior interaction.

show abstract

“…Recent progress on deep neural networks also pushes the development of CTR prediction techniques. A variety of deep CTR prediction models have been proposed and are widely adopted in various large-scale industrial applications such as movie recommender systems, e-commerce systems, and displaying advertisement platforms [10,14,19,23,29,39,40].…”

Section: Introductionmentioning

confidence: 99%

Adversarial Gradient Driven Exploration for Deep Click-Through Rate Prediction

Wu¹,

Chan²,

Bian³

et al. 2021

Preprint

View full text Add to dashboard Cite

Nowadays, data-driven deep neural models have already shown remarkable progress on Click-through Rate (CTR) prediction. Unfortunately, the effectiveness of such models may fail when there are insufficient data. To handle this issue, researchers often adopt exploration strategies to examine items based on the estimated reward, e.g., UCB or Thompson Sampling. In the context of Exploitationand-Exploration for CTR prediction, recent studies have attempted to utilize the prediction uncertainty along with model prediction as the reward score. However, we argue that such an approach may make the final ranking score deviate from the original distribution, and thereby affect model performance in the online system. In this paper, we propose a novel exploration method called Adversarial Gradient Driven Exploration (AGE). Specifically, we propose a Pseudo-Exploration Module to simulate the gradient updating process, which can approximate the influence of the samples of to-be-explored items for the model. In addition, for better exploration efficiency, we propose an Dynamic Threshold Unit to eliminate the effects of those samples with low potential CTR. The effectiveness of our approach was demonstrated on an open-access academic dataset. Meanwhile, AGE has also been deployed in a real-world display advertising platform and all online metrics have been significantly improved.

show abstract

Deep Session Interest Network for Click-Through Rate Prediction

Cited by 51 publications

References 17 publications

Neighbour Interaction based Click-Through Rate Prediction via Graph-masked Transformer

Neighbour Interaction based Click-Through Rate Prediction via Graph-masked Transformer

AskMe: Joint Individual-level and Community-level Behavior Interaction for Question Recommendation

Adversarial Gradient Driven Exploration for Deep Click-Through Rate Prediction

Contact Info

Product

Resources

About