Intention Nets: Psychology-Inspired User Choice Behavior Modeling for Next-Basket Prediction

Abstract

Human behaviors are complex, which are often observed as a sequence of heterogeneous actions. In this paper, we take user choices for shopping baskets as a typical case to study the complexity of user behaviors. Most of existing approaches often model user behaviors in a mechanical way, namely treating a user action sequence as homogeneous sequential data, such as hourly temperatures, which fails to consider the complexity in user behaviors. In fact, users' choices are driven by certain underlying intentions (e.g., feeding the baby or relieving pain) according to Psychological theories. Moreover, the durations of intentions to drive user actions are quite different; some of them may be persistent while others may be transient. According to Psychological theories, we develop a hierarchical framework to describe the goal, intentions and action sequences, based on which, we design Intention Nets (IntNet). In IntNet, multiple Action Chain Nets are constructed to model the user actions driven by different intentions, and a specially designed Persistent-Transient Intention Unit models the different intention durations. We apply the IntNet to next-basket prediction, a recent challenging task in recommender systems. Extensive experiments on real-world datasets show the superiority of our Psychology-inspired model IntNet over the state-of-the-art approaches.

Reinforcement Learning-enhanced Shared-account Cross-domain Sequential Recommendation

“…To be more specific, they first integrate different types of preferences into clusters, and then perform cluster-aware and query-aware graph convolutional propagation on the constructed graph. However, none of the GCNbased methods can be directly applied to SCSR as they either fail to capture the critical sequential information [48] or focus on the recommendation problem in a single domain.…”

“… 48. 22.47 44.18 61.80 62.47 66.59 73.23 13.76 14.16 16.88 18.86 16.88 17.18 20.89 23.64 HGRU4REC 13.76 16.14 22.55 47.98 60.58 63.31 68.00 73.24 13.75 14.14 16.96 20.81 17.04 17.35 20.92 23.64 NAIS 11.45 13.24 19.80 40.17 51.52 54.47 59.30 67.41 10.55 12.57 14.03 16.02 13.29 15.99 14.51 19.82 Time-LSTM 11.19 13.27 20.17 41.45 59.73 60.42 64.66 71.59 11.16 11.60 13.91 18.35 13.74 14.13 17.35 21.17 TGSRec 13.95 15.73 19.91 41.80 50.32 53.40 58.91 67.22 11.91 14.00 14.59 18.44 15.74 16.63 14.57 19.93 π-Net 15.36 17.52 25.13 47.08 60.37 61.74 67.00 74.17 16.24 16.56 18.54 21.87 20.38 20.58 22.44 23.79 PSJNet 15.37 17.56 24.80 46.68 61.89 62.63 66.86 74.14 11.25 13.58 16.25 18.14 15.52 17.30 16.67 19.30 DA-GCN 35.63 37.27 51.35 66.93 59.78 60.55 75.39 82.37 20.09 20.19 22.93 23.90 21.35 21.39 23.93 24.25 TiDA-GCN 38.66 † 40.23 † 54.11 † 68.98 † 63.58 † 65.37 † 76.37 † 83.58 † 20.91 † 21.23 † 23.55 † 24.33 21.88 22.21 † 24.69 † 24.82 Bold value represents the best result of the compared methods in terms of the corresponding metric.…”

Sequence-Based Explainable Hybrid Song Recommendation

“…However, not all of them use them for recommendation with user preferences. In fact, some are session-based CF models that predict the next interaction ( Hidasi et al, 2015 ; Tan et al, 2016 ; Wu et al, 2016 ; Hidasi and Karatzoglou, 2018 ; Yuan et al, 2020 ), or basket of interactions ( Yu et al, 2016 ; Wang Z. et al, 2018 ; Wang et al, 2020a ; Wang et al, 2020b ), in a sequence of interactions regardless of the user’s personal preferences. Similarly, other approaches relied on self-attention networks ( Vaswani et al, 2017 ) to predict the next item recommendation given a sequence of consecutive interactions ( Kang and McAuley, 2018 ; Sun et al, 2019 ; Li et al, 2020 ; Tan et al, 2021 ).…”

On the Exploitation of Blockchain for Distributed File Storage

“…Various recent studies have explored and carried out evaluations for distributed file storage systems [2,6,[19][20][21][22]. Some literatures [2,7,8,[22][23][24][25][26] proposed and evaluated redundancy management strategies. Among these, replication and erasure codes are compared in bandwidth and reliability trade-off in literatures [2,7,8].…”