Bundle Recommendation and Generation With Graph Neural Networks

Jianxin Chang, Chen Gao, Xiangnan He, Depeng Jin, Yong Li

Abstract: Bundle recommendation aims to recommend a bundle of items for a user to consume as a whole. Related work can be divided into two categories: 1) to recommend the platform's prebuilt bundles to users; 2) generate personalized bundles for users. In this work, we propose two graph neural network models, a BGCN model (short for <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"/> <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Bundle</i> <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Graph</i> <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Convolutional</i> <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Network</i> ) for prebuilt bundle recommendation, and a BGGN model (short for <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"/> <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Bundle</i> <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Graph</i> <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Generation</i> <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Network</i> ) for personalized bundle generation. First, BGCN unifies the user-item interaction, the user-bundle interaction and the bundle-item affiliation into a heterogeneous graph. With item nodes as the bridge, graph convolutional propagation between user and bundle nodes makes the learned representations capture the item-level semantics. Second, BGGN re-constructs bundles into graphs based on the item co-occurrence pattern and the user's supervision signal. The complex and high-order item-item relationships in the bundle graph are explicitly modeled through graph generation. Empirical results demonstrate the substantial performance gains of BGCN and BGGN, which outperforms the state-of-the-art baselines by 10.77% to 23.18% and 20.90% to 64.52%, respectively. We have released the datasets and codes at this link: <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/cjx0525/BGCN</uri> .