Fine-tune BERT for E-commerce Non-Default Search Ranking

TL;DR

This paper presents a two-stage ranking method for e-commerce search that combines keyword matching and BERT-based classification, significantly improving relevance in non-default ranking scenarios.

Contribution

It introduces a novel two-stage ranking scheme utilizing BERT fine-tuning and parallel GPU prediction, achieving top results in a data challenge.

Findings

Won 1st place in supervised phase

Achieved 2nd place in final phase

Improved relevance of non-default rankings

Abstract

The quality of non-default ranking on e-commerce platforms, such as based on ascending item price or descending historical sales volume, often suffers from acute relevance problems, since the irrelevant items are much easier to be exposed at the top of the ranking results. In this work, we propose a two-stage ranking scheme, which first recalls wide range of candidate items through refined query/title keyword matching, and then classifies the recalled items using BERT-Large fine-tuned on human label data. We also implemented parallel prediction on multiple GPU hosts and a C++ tokenization custom op of Tensorflow. In this data challenge, our model won the 1st place in the supervised phase (based on overall F1 score) and 2nd place in the final phase (based on average per query F1 score).