Abstract:E-commerce app users exhibit behaviors that are inherently logically consistent. A series of multi-scenario user behaviors interconnect to form the scene-level all-domain user moveline, which ultimately reveals the user's true intention. Traditional CTR prediction methods typically focus on the item-level interaction between the target item and the historically interacted items. However, the scene-level interaction between the target item and the user moveline remains underexplored. There are two challenges when modeling the interaction with preceding all-domain user moveline: (i) Heterogeneity between items and scenes: Unlike traditional user behavior sequences that utilize items as carriers, the user moveline utilizes scenes as carriers. The heterogeneity between items and scenes complicates the process of aligning interactions within a unified representation space. (ii) Temporal misalignment of linked scene-level and item-level behaviors: In the preceding user moveline with a fixed sampling length, certain critical scene-level behaviors are closely linked to subsequent item-level behaviors. However, it is impossible to establish a complete temporal alignment that clearly identifies which specific scene-level behaviors correspond to which item-level behaviors. To address these challenges and pioneer modeling user intent from the perspective of the all-domain moveline, we propose All-domain Moveline Evolution Network (AMEN). AMEN not only transfers interactions between items and scenes to homogeneous representation spaces, but also introduces a Temporal Sequential Pairwise (TSP) mechanism to understand the nuanced associations between scene-level and item-level behaviors, ensuring that the all-domain user moveline differentially influences CTR predictions for user's favored and unfavored items. Online A/B testing demonstrates that our method achieves a +11.6% increase in CTCVR.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the accuracy of click - through rate (CTR) prediction in e - commerce recommendation systems, especially how to better understand users' intentions from the perspective of all - domain user moveline at the scenario level. Traditional CTR prediction methods mainly focus on item - level interactions, that is, the interactions between the target item and historical interaction items, while ignoring the scenario - level interactions between the target item and user movelines. Specifically, the paper aims to address the following two challenges: 1. **Heterogeneity between items and scenarios**: - Traditional user behavior sequences usually take items as the main carriers, while the user movelines that can be captured in the all - domain take scenarios as the main carriers. This difference makes it complicated to align these interactions in a unified representation space. - The paper converts the interactions between items and scenarios into a homogeneous representation space by exploring two paths - item - to - scene and scene - to - item. 2. **Temporal misalignment of linked scenario - level and item - level behaviors**: - In user movelines with a fixed sampling length, some key scenario - level behaviors are closely related to subsequent item - level behaviors, but it is difficult to establish a complete temporal alignment relationship and clarify which specific scenario - level behaviors correspond to which item - level behaviors. - For this reason, the paper introduces the Temporal Sequential Pairwise (TSP) mechanism to capture key temporal dynamics and accurately evaluate the impact of trajectory state changes. To solve these problems, the paper proposes a new CTR prediction model - All - domain Moveline Evolution Network (AMEN). AMEN not only converts the interactions between items and scenarios into a homogeneous representation space, but also understands the subtle associations between scenario - level and item - level behaviors through the TSP mechanism, ensuring that all - domain user movelines have a differential impact on the CTR prediction of users' preferred and non - preferred items. Experimental results show that AMEN performs well in online A/B tests on large - scale real - world data, increasing the Click - Through Conversion Rate after Viewing (CTCVR) by 11.6%. This proves the effectiveness and superiority of AMEN in understanding and predicting users' intentions.

All-domain Moveline Evolution Network for Click-Through Rate Prediction

Click-Through Rate Prediction Algorithm Based on Modeling of Implicit High-Order Feature Importance

Collaborative Contrastive Network for Click-Through Rate Prediction

Deep Situation-Aware Interaction Network for Click-Through Rate Prediction.

ENCODE: Breaking the Trade-Off Between Performance and Efficiency in Long-Term User Behavior Modeling

EXTR: Click-Through Rate Prediction with Externalities in E-Commerce Sponsored Search

Deep Interest Evolution Network for Click-Through Rate Prediction

A Deep Behavior Path Matching Network for Click-Through Rate Prediction

An adaptive hybrid XdeepFM based deep Interest network model for click-through rate prediction system

A disaggregated interest-extraction network for click-through rate prediction

Context-aware cross feature attentive network for click-through rate predictions

End-to-End User Behavior Retrieval in Click-Through RatePrediction Model

Branches, Assemble! Multi-Branch Cooperation Network for Large-Scale Click-Through Rate Prediction at Taobao

Deep Context Interest Network for Click-Through Rate Prediction

Time-aligned Exposure-enhanced Model for Click-Through Rate Prediction

Decision-Making Context Interaction Network for Click-Through Rate Prediction

Neighbour Interaction based Click-Through Rate Prediction via Graph-masked Transformer

Multifeedback Behavior-Based Interest Modeling Network for Adaptive Click-Through Rate Prediction

Practice on Long Sequential User Behavior Modeling for Click-Through Rate Prediction

Context-Aware Modeling Via Simulated Exposure Page for CTR Prediction

DMBIN: A Dual Multi-behavior Interest Network for Click-Through Rate Prediction Via Contrastive Learning