Search papers, labs, and topics across Lattice.
This paper introduces TRUST, a novel framework for session-based recommendation that recalibrates temporal signals by evaluating each observed interval relative to the empirical interval distribution of individual items. The authors demonstrate that traditional methods, which assume uniformity in interval significance across items, overlook critical variations in user interest, leading to suboptimal recommendations. Experimental results across public datasets reveal that TRUST significantly outperforms both temporal and non-temporal baselines, with component-wise ablations confirming that calibrated temporal signals enhance the performance of various recommendation modules.
By recalibrating temporal signals based on item-specific interval distributions, TRUST achieves superior recommendation accuracy, challenging the one-size-fits-all approach in session-based systems.
Temporal signals have been widely used in session-based recommendation to infer user interest. Existing temporal session-based recommenders primarily rely on absolute interval values, implicitly assuming that the same interval carries similar interest signals across items. However, we empirically find that this assumption does not hold: each item has its own interval distribution, so an interval should be interpreted relative to the item it belongs to. Based on this observation, we propose TRUST, a framework that evaluates each observed interval relative to the empirical interval distribution of the corresponding item. Specifically, we propose a score function to guide global neighbor sampling, session graph encoding, and final interest aggregation. Experiments on public datasets show that TRUST consistently improves over representative temporal and non-temporal baselines, and plug-in experiments further show that the proposed scoring function can improve existing temporal session recommenders as a model-agnostic method. Component-wise ablations further show that calibrating the temporal signals within each module, rather than removing the module itself, consistently improves neighbor sampling, session graph encoding, and interest aggregation.
