Transfer Meets Hybrid: A Synthetic Approach for Cross-Domain Collaborative Filtering with Text

Hu, Guangneng; Zhang, Yu; Yang, Qiang

doi:10.1145/3308558.3313543

Cited by 84 publications

(36 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is clear to see that Equation (12) adopts the idea of feature propagation on the graph, which forms improved feature by aggregating the features from A p (e ( ) ) and B p (e ( ) ). In this way, we can regard ( ) as the common features derived for under both A and B .…”

Section: = |N | |N | + |N |mentioning

confidence: 99%

“…When the interaction history is less in domain A, it is natural to consider getting some common knowledge from correlated domain B that includes more data. In recent years, cross-domain collaborative filtering (CDCF) [2,11,12,15,22,28] has attracted increasing research attention. But like every coin has two sides, the correlations between domains make CDCF possible, the differences between domains also render it difficult to transfer knowledge.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Cross Domain Recommendation via Bi-directional Transfer Graph Collaborative Filtering Networks

Liu

et al. 2020

Proceedings of the 29th ACM International Conference on Information &Amp; Knowledge Management

123

View full text Add to dashboard Cite

Data sparsity is a challenge problem that most modern recommender systems are confronted with. By leveraging the knowledge from relevant domains, the cross-domain recommendation technique can be an effective way of alleviating the data sparsity problem. In this paper, we propose a novel Bi-directional Transfer learning method for cross-domain recommendation by using Graph Collaborative Filtering network as the base model (BiTGCF). BiT-GCF not only exploits the high-order connectivity in user-item graph of single domain through a novel feature propagation layer, but also realizes the two-way transfer of knowledge across two domains by using the common user as the bridge. Moreover, distinct from previous cross-domain collaborative filtering methods, BiTGCF fuses users' common features and domain-specific features during transfer. Experimental results on four couple benchmark datasets verify the effectiveness of BiTGCF over state-of-the-art models in terms of bi-directional cross domain recommendation. CCS CONCEPTS • Information systems → Recommender systems; • Computing methodologies → Transfer learning.

show abstract

Section: = |N | |N | + |N |mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Cross Domain Recommendation via Bi-directional Transfer Graph Collaborative Filtering Networks

Liu

et al. 2020

Proceedings of the 29th ACM International Conference on Information &Amp; Knowledge Management

123

View full text Add to dashboard Cite

show abstract

“…Yuan et al [29] encoded each user by an auto-encoder and aligned the user representations by domain adaptation. Hu et al [10] used items on the source domain which have interactions with current user to enhance the representation of the current item on the target domain, and used user reviews to improve the model performance. We can see that transfer learning is easy to achieve since there is user and item overlap.…”

Section: Cross-domain Recommendation Withmentioning

confidence: 99%

Semi-supervised Collaborative Filtering by Text-enhanced Domain Adaptation

Lin

et al. 2020

Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery &Amp; Data Mining

View full text Add to dashboard Cite

Data sparsity is an inherent challenge in the recommender systems, where most of the data is collected from the implicit feedbacks of users. This causes two difficulties in designing effective algorithms: first, the majority of users only have a few interactions with the system and there is no enough data for learning; second, there are no negative samples in the implicit feedbacks and it is a common practice to perform negative sampling to generate negative samples. However, this leads to a consequence that many potential positive samples are mislabeled as negative ones and data sparsity would exacerbate the mislabeling problem. To solve these difficulties, we regard the problem of recommendation on sparse implicit feedbacks as a semi-supervised learning task, and explore domain adaption to solve it. We transfer the knowledge learned from dense data to sparse data and we focus on the most challenging case âĂŤ there is no user or item overlap. In this extreme case, aligning embeddings of two datasets directly is rather sub-optimal since the two latent spaces encode very different information. As such, we adopt domain-invariant textual features as the anchor points to align the latent spaces. To align the embeddings, we extract the textual features for each user and item and feed them into a domain classifier with the embeddings of users and items. The embeddings are trained to puzzle the classifier and textual features are fixed as anchor points. By domain adaptation, the distribution pattern in the source domain is transferred to the target domain. As the target part can be supervised by domain adaptation, we abandon negative sampling in target dataset to avoid label noise.

show abstract

“…Our proposed CD-DNN is different from all the above mentioned cross-domain recommendation methods because we also utilize review text to mine more precise features of users. To the best of our knowledge, there is only one work [38] that adopts review text to improve recommendation accuracy in the cross-domain recommender system. They proposed a Transfer Meeting Hybrid model for cross-domain recommendation with text such as item reviews and article titles end-to-end.…”

Section: Cross-domain Recommendationmentioning

confidence: 99%

A Parallel Deep Neural Network Using Reviews and Item Metadata for Cross-Domain Recommendation

et al. 2020

View full text Add to dashboard Cite

Cross-domain recommendation is an effective technique to alleviate the data sparsity problem in recommender systems by utilizing the information from relevant domains. In this paper, we propose Crossdomain Deep Neural Network (CD-DNN) for the cross-domain recommendation. CD-DNN solves the rating prediction problem by modeling users and items using reviews and item metadata, which jointly learns features of users and items from not only the target domain but also other source domains. Latent factors for users and items are learned by several parallel neural networks, and the relevance of user features and item features is learned by maximizing prediction accuracy. CD-DNN builds a single mapping for user features in the latent space, so that the network for user is optimized together with item features from other domains. Experimental results indicate that the proposed CD-DNN significantly outperforms other state-of-the-art recommendation approaches on four public datasets of Amazon and it alleviates the data sparsity problem by leveraging more data across domains. INDEX TERMS Cross-domain recommendation, convolutional neural networks, rating prediction. NANNAN ZHENG received the Bachelor of Engineering degree from Xiamen University, China, where she is currently pursuing the degree with the Department of Automation. Her current research interests are in deep learning and recommendation systems at the System and Control Center Laboratory, Xiamen University. ZIANG XIONG received the Bachelor of Engineering degree from Xiamen University, China, in 2018, where he is currently pursuing the degree with the Department of Automation. His current research interests are in deep learning and recommendation systems at the System and Control Center Laboratory, Xiamen University. ZHIQIANG HU received the Bachelor of Engineering degree from the Chongqing University of Posts and Telecommunications, China. He is currently pursuing the degree with the Department of Automation, Xiamen University, China. His current research interests are in natural language processing and knowledge graph at the System and

show abstract

Transfer Meets Hybrid: A Synthetic Approach for Cross-Domain Collaborative Filtering with Text

Cited by 84 publications

References 51 publications

Cross Domain Recommendation via Bi-directional Transfer Graph Collaborative Filtering Networks

Cross Domain Recommendation via Bi-directional Transfer Graph Collaborative Filtering Networks

Semi-supervised Collaborative Filtering by Text-enhanced Domain Adaptation

A Parallel Deep Neural Network Using Reviews and Item Metadata for Cross-Domain Recommendation

Contact Info

Product

Resources

About