Outer product enhanced heterogeneous information network embedding for recommendation

He, Yunfei; Zhang, Yiwen; Qi, Lianyong; Yan, Dengcheng; He, Qiang

doi:10.1016/j.eswa.2020.114359

Cited by 24 publications

(5 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Here the outer product is used to produce the matrix related to the feature relationship among the user and item embeddings over the heterogeneous graph [41]. At the end, a rating estimation function for matrix factorization (MF) is obtained.…”

Section: The Review Effected Hrec (Rehrec) Approachmentioning

confidence: 99%

REHREC: Review Effected Heterogeneous Information Network Recommendation System

Khalilzadeh,

Cicekli

2024

IEEE Access

View full text Add to dashboard Cite

Heterogeneous data networks have bunches of rich secret data that assist us in the creation of successful recommendation frameworks. Specifying the meaningful meta-paths helps us to extract the hidden information that exists in a heterogeneous data network. Although user and business nodes are used to specify these meaningful meta-paths, review nodes have not been used. Furthermore, written reviews assist us to obtain informative data about a point of interest in a recommendation system. We use new meta-paths which employs review nodes in order to extract the hidden information in heterogeneous data networks. In this work, it is tried to unify rich written review texts with the heterogeneous information network and their effects in recommendation systems are analyzed. Our experiments show that the review texts improve the recommendation system when meaningful meta-paths are chosen.

show abstract

Section: The Review Effected Hrec (Rehrec) Approachmentioning

confidence: 99%

REHREC: Review Effected Heterogeneous Information Network Recommendation System

Khalilzadeh,

Cicekli

2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Therefore, there are many variant models developed based on GNN, such as recurrent GNN [15], convolutional GNN [16], [17], graph auto-encoders [18], [19], and spatial-temporal GNN [20]. Graph attention network [21] is one of the modern models widely applied in fields such as link prediction [22]- [24], node classification [25], node clustering [26], recommendation system [27], [28], information diffusion [29], and in this paper, we apply this technique to resolve the anchor link prediction problem.…”

Section:  Issn: 2302-9285mentioning

confidence: 99%

An approach of anchor link prediction using graph attention mechanism

Phạm

Toai

et al. 2022

Bulletin EEI

View full text Add to dashboard Cite

Nowadays social networks such as Twitter, LinkedIn, and Facebook are a popular and necessary platform. It is considered a miniature of an actual social network because of its advantages in connecting and sharing information between users. The analysis of data on online social networks has become a field that has attracted a lot of attention from the research community and anchor link prediction is one of the main research directions in this field. Depending on demand, a user can simultaneously participate in many different online social networks, anchor link prediction is a kind of task that determines the identity of a user on many different social networks. In this article, we proposed an algorithm that determines missing/future anchor links between users from two different online social networks. Our algorithm utilizes the graph attention technique to represent the source and target network into the low-dimension embedding spaces, we then apply the canonical correlation analysis to recline their embeddings into same latent spaces for final prediction.

show abstract

“…For example, the matrix factorization is utilized in HeteRec [29], FMG [?, fmg]nd HueRec [26] to accomplish this step, while the HNAFM [1] and NeuACF [17] adopt MLP to extract deeper information of users and items according to their initialized features in each metapath. Besides, the HERec [18] and HopRec [6] capture the graph structure in corresponding meta-path with the help of DeepWalk [13]. Generally, these methods do not take into account both graph structure and features simultaneously.…”

Section: Hin-based Recommendation Systemsmentioning

confidence: 99%

“…This process merges multi aspects auxiliary information into one representation which can used to recommendation. However, meta-paths in the HIN can be formed as graphs, and the extracting stage in existing RSs [1] [17] [18] [6] based on HIN usually focus on the single graph structure or user/item feature, while not take both them into account. The lack of graph structure information or features can cause the embedding extracted from each meta-path has an insufficient expression ability and further decrease the performance of RSs based on HIN.…”

Section: Introductionmentioning

confidence: 99%

Heterogeneous Information Network-based Interest Composition with Graph Neural Network for Recommendation

Yan,

Xie,

Zhang

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Recommendation systems (RSs) are skilled at capturing users' preferences according to their history interactions with items, while the RSs usually suffer from the sparsity of user-item interactions. Thus, various auxiliary information is introduced to alleviate this problem. Due to powerful ability in modeling auxiliary information, the heterogeneous information network (HIN) is widely applied to the RSs. However, in existing methods, the process of information extraction from various meta-paths takes no consideration of graph structure and user/item features simultaneously. Moreover, existing models usually fuse the information from various meta-paths through simply weighted summation, while ingore the interest compositions intra-and inter-meta-paths which is capable of applying abundant high-order composition interests to RSs. Therefore, we propose a HIN-based Interest Compositions model with graph neural network for Recommendation (short for HicRec). Above all, our model learns users and items representations from various graphs corresponding to the meta-paths with the help of the graph convolution network (GCN). Then, the representations of users and items are transformed into users' interests on items. Lastly, the interests intra-and inter-meta-paths are composed and applied to recommendation. Extensive experiments are conducted on three real-world datasets and the results show that the HicRec outperforms various baselines.

show abstract

Outer product enhanced heterogeneous information network embedding for recommendation

Cited by 24 publications

References 37 publications

REHREC: Review Effected Heterogeneous Information Network Recommendation System

REHREC: Review Effected Heterogeneous Information Network Recommendation System

An approach of anchor link prediction using graph attention mechanism

Heterogeneous Information Network-based Interest Composition with Graph Neural Network for Recommendation

Contact Info

Product

Resources

About