TIM: Temporal Interaction Model in Notification System

TL;DR

The paper addresses the challenge of optimally timing mobile notifications by modeling long-term user behavior and introducing a holistic scheduling approach for multiple notifications per day. It proposes TIM, which uses a Temporal Attention Unit (TAU) to extract work/rest day–dependent patterns from long-term interaction sequences and predict per-slot CTRs $p_k$, followed by a sending strategy that allocates notifications proportionally via $n_i = q\cdot\frac{p_i}{\|\mathbf{p}\|_1}$ and a cumulative-CTR trigger. Offline experiments on Kuaishou data show TIM outperforms strong baselines, with ablations confirming the benefits of RoPE and Day-Type Embeddings in capturing daily patterns. Online A/B tests demonstrate TIM increases CTR and modestly boosts DAU and watch time while keeping notification volume and switch-off rates neutral, indicating practical engagement gains without disturbing users. Overall, TIM provides a scalable framework for holistic notification scheduling with real-world impact and potential extensions to real-time and multi-objective optimization.

Abstract

Modern mobile applications heavily rely on the notification system to acquire daily active users and enhance user engagement. Being able to proactively reach users, the system has to decide when to send notifications to users. Although many researchers have studied optimizing the timing of sending notifications, they only utilized users' contextual features, without modeling users' behavior patterns. Additionally, these efforts only focus on individual notifications, and there is a lack of studies on optimizing the holistic timing of multiple notifications within a period. To bridge these gaps, we propose the Temporal Interaction Model (TIM), which models users' behavior patterns by estimating CTR in every time slot over a day in our short video application Kuaishou. TIM leverages long-term user historical interaction sequence features such as notification receipts, clicks, watch time and effective views, and employs a temporal attention unit (TAU) to extract user behavior patterns. Moreover, we provide an elegant strategy of holistic notifications send time control to improve user engagement while minimizing disruption. We evaluate the effectiveness of TIM through offline experiments and online A/B tests. The results indicate that TIM is a reliable tool for forecasting user behavior, leading to a remarkable enhancement in user engagement without causing undue disturbance.

Figures (2)

• Figure 1: Overview of TIM, with TAU as the central structure for extracting user behavioral patterns. TAU extracts the feature of the target date from the day dimension and then fuses the features in the slot dimension.
• Figure 2: Attention scores in TIM (a) and $\text{TIM}_{R}$ (b).