Using Large Pretrained Language Models for Answering User Queries from Product Specifications
Authors:
Kalyani Roy,
Smit Shah,
Nithish Pai,
Jaidam Ramtej,
Prajit Prashant Nadkarn,
Jyotirmoy Banerjee,
Pawan Goyal,
Surender Kumar
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…
▽ More
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.
△ Less
Submitted 29 May, 2020;
originally announced May 2020.
