A Survey on E-Commerce Learning to Rank

TL;DR

This paper addresses the challenge of ranking relevance in e-commerce search by surveying learning-to-rank (LTR) methodologies and providing an empirical comparison on a real dataset. It organizes LTR approaches into pointwise, pairwise, and listwise frameworks, and investigates representation learning strategies including BERT/eBERT and spherical text embeddings. The study highlights practical findings: gradient boosting-based pointwise methods, LambdaMART-based pairwise methods, and ListMLE-based listwise methods generally perform best, with e-commerce-specific considerations influencing method choice. The work underscores the scarcity of public e-commerce datasets and aims to guide practitioners toward effective, dataset-informed LTR selections while signaling future directions in deep learning for this domain.

Abstract

In e-commerce, ranking the search results based on users' preference is the most important task. Commercial e-commerce platforms, such as, Amazon, Alibaba, eBay, Walmart, etc. perform extensive and relentless research to perfect their search result ranking algorithms because the quality of ranking drives a user's decision to purchase or not to purchase an item, directly affecting the profitability of the e-commerce platform. In such a commercial platforms, for optimizing search result ranking numerous features are considered, which emerge from relevance, personalization, seller's reputation and paid promotion. To maintain their competitive advantage in the market, the platforms do no publish their core ranking algorithms, so it is difficult to know which of the algorithms or which of the features is the most effective for finding the most optimal search result ranking in e-commerce. No extensive surveys of ranking to rank in the e-commerce domain is also not yet published. In this work, we survey the existing e-commerce learning to rank algorithms. Besides, we also compare these algorithms based on query relevance criterion on a large real-life e-commerce dataset and provide a quantitative analysis. To the best of our knowledge this is the first such survey which include an experimental comparison among various learning to rank algorithms.

Figures (3)

• Figure 1: Illustration of a LTR model to perform e-commerce product ranking task.
• Figure 2: Illustration of a same product appearing multiple times of a single query.
• Figure 3: Query-Product Similarity Value Distribution of the Train, Dev, and Test Partitions of the e-commerce Dataset