Large Scale Question Paraphrase Retrieval with Smoothed Deep Metric Learning
This addresses noisy question reformulations for community and spoken language question answering systems, but is incremental as it improves an existing loss function for a specific task.
The paper tackles the problem of retrieving equivalent questions with noisy labels in Question Paraphrase Retrieval (QPR) by proposing a smoothed deep metric loss (SDML), which significantly outperforms the standard triplet loss in experiments on two datasets.
The goal of a Question Paraphrase Retrieval (QPR) system is to retrieve equivalent questions that result in the same answer as the original question. Such a system can be used to understand and answer rare and noisy reformulations of common questions by mapping them to a set of canonical forms. This has large-scale applications for community Question Answering (cQA) and open-domain spoken language question answering systems. In this paper we describe a new QPR system implemented as a Neural Information Retrieval (NIR) system consisting of a neural network sentence encoder and an approximate k-Nearest Neighbour index for efficient vector retrieval. We also describe our mechanism to generate an annotated dataset for question paraphrase retrieval experiments automatically from question-answer logs via distant supervision. We show that the standard loss function in NIR, triplet loss, does not perform well with noisy labels. We propose smoothed deep metric loss (SDML) and with our experiments on two QPR datasets we show that it significantly outperforms triplet loss in the noisy label setting.