Using Large Pretrained Language Models for Answering User Queries from Product Specifications

TL;DR

This paper explores using large pretrained language models like XLNet and BERT to automatically identify relevant product specifications for answering user queries on e-commerce sites, improving response accuracy across different product categories.

Contribution

It introduces a method to create training data automatically and demonstrates that pretrained models outperform Siamese models in this task, with cross-vertical generalization.

Findings

Pretrained models outperform Siamese models in accuracy.

Models trained on one vertical generalize well to others.

Automatic dataset creation is effective for this task.

Abstract

While buying a product from the e-commerce websites, customers generally have a plethora of questions. From the perspective of both the e-commerce service provider as well as the customers, there must be an effective question answering system to provide immediate answers to the user queries. While certain questions can only be answered after using the product, there are many questions which can be answered from the product specification itself. Our work takes a first step in this direction by finding out the relevant product specifications, that can help answering the user questions. We propose an approach to automatically create a training dataset for this problem. We utilize recently proposed XLNet and BERT architectures for this problem and find that they provide much better performance than the Siamese model, previously applied for this problem. Our model gives a good performance even when trained on one vertical and tested across different verticals.