Learning to Generate Questions by LearningWhat not to Generate

Abstract

Automatic question generation is an important technique that can improve the training of question answering, help chatbots to start or continue a conversation with humans, and provide assessment materials for educational purposes. Existing neural question generation models are not sufficient mainly due to their inability to properly model the process of how each word in the question is selected, i.e., whether repeating the given passage or being generated from a vocabulary. In this paper, we propose our Clue Guided Copy Network for Question Generation (CGC-QG), which is a sequence-to-sequence generative model with copying mechanism, yet employing a variety of novel components and techniques to boost the performance of question generation. In CGC-QG, we design a multi-task labeling strategy to identify whether a question word should be copied from the input passage or be generated instead, guiding the model to learn the accurate boundaries between copying and generation. Furthermore, our input passage encoder takes as input, among a diverse range of other features, the prediction made by a clue word predictor, which helps identify whether each word in the input passage is a potential clue to be copied into the target question. The clue word predictor is designed based on a novel application of Graph Convolutional Networks onto a syntactic dependency tree representation of each passage, thus being able to predict clue words only based on their context in the passage and their relative positions to the answer in the tree. We jointly train the clue prediction as well as question generation with multi-task learning and a number of practical strategies to reduce the complexity. Extensive evaluations show that our model significantly improves the performance of question generation and out-performs all previous state-of-the-art neural question generation models by a substantial margin.

Asking Questions the Human Way: Scalable Question-Answer Generation from Text Corpus

“…The copy probability of each input word is given by the attention weights in Equation (10). It has been reported that the generated words in a target question are usually from frequent words, while the majority of lowfrequency words in the long tail are copied from the input instead of generated [27]. Therefore, we reduce the vocabulary size to be the top N V high-frequency words at both the encoder and the decoder, where N V is a predefined threshold that varies among different datasets.…”

“…• NQG-Knowledge [16], DLPH [12]: auxiliary-informationenhanced question generation models with extra inputs such as knowledge or difficulty. • Self-training-EE [38], BERT-QG-QAP [51], NQG-LM [55], CGC-QG [27] and QType-Predict [56]: multi-task question generation models with auxiliary tasks such as question answering, language modeling, question type prediction and so on.…”

Diversify Question Generation with Continuous Content Selectors and Question Type Modeling

“…Automatic question generation has attracted an increasing attention from the natural language gen-eration community in recent years, which is reflected in newly published datasets (Zhou et al, 2017;Chen et al, 2018) and sophisticated techniques (Du et al, 2017;Liu et al, 2019). Traditional methods are mainly rule-based, where they first transform the source information into syntactic representation and then use templates to generate related questions (Heilman, 2011).…”

Fine-Grained Privacy Detection with Graph-Regularized Hierarchical Attentive Representation Learning

“…As an extension of the convolutional network, graph convolutional network, introduced by Bruna et al in [6], works on exploiting the adjacency matrix or the Laplacian matrix that characterizes the graph structure, and, hence, capturing the correlations among different nodes. Due to its powerful capability in exploring the correlation propagation between nodes, recent years have witnessed increasing research attention from both the natural language processing domain [3,36,42,49] and the computer vision domain [9,20,25,26] has been paid to the graph convolutional network. For example, Kipf and Welling [36] introduced a graph-based semi-supervised learning framework for node classifications, where the label information is smoothed over the graph via a Laplacian regularization term in the loss function.…”