EENMF: An End-to-End Neural Matching Framework for E-Commerce Sponsored Search

TL;DR

This paper introduces EENMF, an end-to-end neural framework for e-commerce sponsored search that improves ad retrieval and pre-ranking by integrating deep matching models, leading to significant performance gains in real-world traffic.

Contribution

The paper presents a novel end-to-end neural matching framework that unifies ad retrieval and pre-ranking, optimizing the entire process for better effectiveness and efficiency.

Findings

Significant performance improvement over baseline in real traffic.

Effective integration of vector-based retrieval with neural pre-ranking.

Optimized pointwise cross-entropy loss enhances ranking accuracy.

Abstract

E-commerce sponsored search contributes an important part of revenue for the e-commerce company. In consideration of effectiveness and efficiency, a large-scale sponsored search system commonly adopts a multi-stage architecture. We name these stages as ad retrieval, ad pre-ranking and ad ranking. Ad retrieval and ad pre-ranking are collectively referred to as ad matching in this paper. We propose an end-to-end neural matching framework (EENMF) to model two tasks---vector-based ad retrieval and neural networks based ad pre-ranking. Under the deep matching framework, vector-based ad retrieval harnesses user recent behavior sequence to retrieve relevant ad candidates without the constraint of keyword bidding. Simultaneously, the deep model is employed to perform the global pre-ranking of ad candidates from multiple retrieval paths effectively and efficiently. Besides, the proposed model tries to optimize the pointwise cross-entropy loss which is consistent with the objective of predict models in the ranking stage. We conduct extensive evaluation to validate the performance of the proposed framework. In the real traffic of a large-scale e-commerce sponsored search, the proposed approach significantly outperforms the baseline.