CombiGCN: An effective GCN model for Recommender System
Loc Tan Nguyen, Tin T. Tran · 2025
Read CombiGCN: An effective GCN model for Recommender System by Loc Tan Nguyen, Tin T. Tran (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-21471.txt.
Introduction
Loc Nguyen and Tin Tran present CombiGCN, a recommendation model built on graph neural networks. Recommenders guess what a user will like from past clicks and purchases, and graph methods spread that signal through the web of users and items. The authors note a blind spot: designs update a user's profile only from the items they touched, ignoring how alike two users are. Their model adds a second graph that links users by the similarity of their histories.
Item profiles travel only along the user-item graph. User profiles travel along both graphs at once, and the two streams are merged layer by layer with a weighted sum. The team tested the design against leading models on three real-world data sets. They frame the user-user links as the missing collaborative signal that lets the model borrow taste from peers, not from shared items.
Catalog record
| CREATOR | Loc Tan Nguyen, Tin T. Tran |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 19 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-21471
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
CombiGCN: An effective GCN model for Recommender System Loc Tan Nguyen, Tin T. Tran Abstract Graph Neural Networks (GNNs) have opened up a potential line of research for collaborative filtering (CF). The key power of GNNs is based on injecting collaborative signal into user and item embeddings which will contain information about user-item interactions after that. However, there are still some unsatisfactory points for a CF model that GNNs could have done better. The way in which the collaborative signal are extracted through an implicit feedback matrix that is essentially built on top of the message-passing architecture of GNNs, and it only helps to update the embedding based on the value of the items (or users) embeddings neighboring. By identifying the similarity weight of users through their interaction history, a key concept of CF, we endeavor to build a user-user weighted connection graph based on their similarity weight. In this study, we propose a recommendation framework, CombiGCN, in which item embeddings are only linearly propagated on the user-item interaction graph, while user embeddings are propagated simultaneously on both the user-user weighted connection graph and user-item interaction graph graphs with Light Graph Convolution (LGC) and combined in a simpler method by using the weighted sum of the embeddings for each layer. We also conducted experiments comparing CombiGCN with several state-of-the-art models on three real-world datasets. † † footnotetext: * Corresponding author: Tin T. Tran, [email protected] 1 Introduction Recommendation systems play an important role in online businesses because of the economic benefits they bring by suggesting suitable products or services to customers. That motivation has driven research to improve algorithms to offer powerful recommendation engines, typically collaborative filtering (CF). Concurrent with the rise of deep learning, especially the use of GNNs to learn representations of users and items (as known as embeddings), many recent studies have focused on enriching embeddings by encoding them with collaborative signals, which carry information about user-item interactions [ 2 , 5 , 3 , 4 , 6 ] . These signals are extracted through the message-passing architecture of GNNs. More specifically, considering a user u as a node in the graph whose embedding is e_{u} , at each propagation time this user node will adjust its embedding by aggregating all embeddings of neighboring items. During the aggregation progress, each embedding e_{i} from a neighboring node item i will be multiplied by a coefficient p_{ui}=1/\sqrt{|\mathcal{N}_{u}||\mathcal{N}_{i}|} , where \mathcal{N}_{u} and \mathcal{N}_{i} denote the first-hop neighbors of user u and item i , so the updated embedding value e_{u} not only carries information about neighboring items, it also reflects the mutual importance between user u and item i through the p_{ui} coefficient. However, GNN models only help each user or item embedding in the user-item interaction graph to be similar to neighboring nodes without regard to the weights of the links between nodes during the entire propagation process. There has been some research on adding weights to the embedding encoding process, such as [ 5 , 6 ] . These studies construct a user-user graph where each connection between two users is the number of items shared by them. Aiming at addressing this problem, we have normalized the user-user graph based on Jaccard similarity and integrated these weights, which improves the quality of the extracted collaborative signal over each propagation and produces satisfactory embeddings for CF. Combining embedding propagated from two graphs has also been conducted through many studies [ 6 , 4 , 5 ] . SocialLGN has proven that their proposed graph fusion operation is a state-of-the-art combination embeddings from the two graph method [ 4 ] . Their results are more accurate than results from graph fusion operations based on GCN methods [ 7 ] or GraphSage [ 8 ] . In this paper, we propose a model named CombiGCN based on Light Graph Convolution (LGC) [ 3 ] to propagate user and item embeddings on the user-item interaction graph; in the meantime, user embedding is also propagated on the user-user weighted connection graph. To fuse two user embeddings obtained after each propagation into an integrated embedding, instead of using the fusion graph operation of SocialLGN, we simply use the weighted sum of the embeddings. We demonstrate the superior performance of CombiGCN by comparing it with state-of-the-art recommendation models on three real-world datasets that we preprocessed to avoid cold-start and over-fitting. 2 Related Work 2.1 Graph Convolution Networks Due to its superior ability to model graph-structured data, Graph Neural Networks (GNNs) have become the state-of-the-art technology for recommendation systems. A graph convolution network (GCN) is a special type of GNNs that uses convolution aggregations. Spatial GCNs based on 1-hop neighbor propagation and in each layer GCN neighborhood aggregations are required and thus the computational cost is significantly reduced. In the recommended context, spatial GCN contains NGCF [ 2 ] , WiGCN [ 5 ] and GCMC [ 9 ] . A recent study [ 3 ] developed Light Graph Convolution (LGC) based on the GCN architecture but eliminates trainable matrices and non-linear activation functions in GCN. The experiment reported in [ 3 ] also shows that LGC outperform GCN. Today’s most modern CF models [ 3 , 4 , 6 ] also use LGC instead of traditional GCN. 2.2 Multi-layer Perceptron and Backpropagation In machine learning models involving Neural Networks like GNNs, learning the trainable parameters includes forward-propagation and back-propagation. The forward-propagation stage calculates the embedding value of each node in the neural network with trainable parameters. In matrix form, these parameters include the trainable matrices W^{k} and embeddings E^{k} in the k -th layer. In addition, non-linear activation functions such as ReLU and tanh are also applied to the results. The calculations in forward-propagation produce the prediction result in the last layer. To train the model, an optimization function to this result and propagate back to adjust the trainable parameters [ E,W ]. When training neural networks, forward-propagation and back-propagation depend on each other. For the recommendation problem using GNNs also follows this rule. However, a recent study [ 3 ] has demonstrated the redundancy of the trainable matrices W^{k} and non-linear activation functions in recommendation models, the reason being that the embeddings mapped from user and item IDs are not many features, so using too many trainable parameters makes the model heavy and ineffective. 3 Our proposed method We proposed our model, which includes a method to pre-process data set and the CombiGCN model, in this chapter. CombiGCN will explores the user and item interaction and weighted similarity matrix as the input, and make prediction at the output as recommendations. The overview of CombiGCN model is illustrated in Figure 1 . 3.1 Pre-processing data Algorithm 1 Inference for data pre-processing \mathcal{U}\times\mathcal{I} \triangleright Interaction between users and items in original dataset r \triangleright ratio between users and items you want to obtain R=U\times I\subseteq\mathcal{U}\times\mathcal{I} for each item i\in\mathcal{I} do set_{i}\leftarrow\textrm{list of users that interact with item }i len_{i}=\textrm{total number of users in }set_{i} end for \widetilde{m}\leftarrow\textrm{cardinality of set }I I_{\textrm{c}}\leftarrow\textrm{m items have highest }len_{i} \widetilde{n}\leftarrow\widetilde{m}\times r for each user u\in\mathcal{U} do set_{u}\leftarrow\textrm{list of items interacted by user }u sim_{u}=\textrm{Jaccard distance between}I_{\textrm{c}}\textrm{and }set_{u} end for U_{\textrm{c}}\leftarrow\widetilde{n}\textrm{ users have highest }sim_{u} U\leftarrow n\textrm{ users from }\widetilde{n}\textrm{ users of