Open-domain question answering has been used in a wide range of applications, such as web search and enterprise search, which usually takes clean texts extracted from various formats of documents (e.g., web pages, PDFs, or Word documents) as the information source. However, designing different text extraction approaches is time-consuming and not scalable. In order to reduce human cost and improve the scalability of QA systems, we propose and study an Open-domainDocument Visual Question Answering (Open-domain DocVQA) task, which requires answering questions based on a collection of document images directly instead of only document texts, utilizing layouts and visual features additionally. Towards this end, we introduce the first Chinese Open-domain DocVQA dataset called DuReadervis, containing about 15K question-answering pairs and 158K document images from the Baidu search engine. There are three main challenges in DuReadervis: (1) long document understanding, (2) noisy texts, and (3) multi-span answer extraction. The extensive experiments demonstrate that the dataset is challenging. Additionally, we propose a simple approach that incorporates the layout and visual features, and the experimental results show the effectiveness of the proposed approach. The dataset and code will be publicly available at https://github.com/baidu/DuReader/tree/master/DuReader-vis.
In this paper, we present DuReader-retrieval, a large-scale Chinese dataset for passage retrieval. DuReader-retrieval contains more than 90K queries and over 8M unique passages from a commercial search engine. To alleviate the shortcomings of other datasets and ensure the quality of our benchmark, we (1) reduce the false negatives in development and test sets by manually annotating results pooled from multiple retrievers, and (2) remove the training queries that are semantically similar to the development and testing queries. Additionally, we provide two out-of-domain testing sets for cross-domain evaluation, as well as a set of human translated queries for for cross-lingual retrieval evaluation. The experiments demonstrate that DuReader-retrieval is challenging and a number of problems remain unsolved, such as the salient phrase mismatch and the syntactic mismatch between queries and paragraphs. These experiments also show that dense retrievers do not generalize well across domains, and cross-lingual retrieval is essentially challenging. DuReader-retrieval is publicly available at https://github.com/baidu/DuReader/tree/master/DuReader-Retrieval.
Machine reading comprehension (MRC) is a crucial task in natural language processing and has achieved remarkable advancements. However, most of the neural MRC models are still far from robust and fail to generalize well in real-world applications. In order to comprehensively verify the robustness and generalization of MRC models, we introduce a real-world Chinese dataset – DuReader_robust . It is designed to evaluate the MRC models from three aspects: over-sensitivity, over-stability and generalization. Comparing to previous work, the instances in DuReader_robust are natural texts, rather than the altered unnatural texts. It presents the challenges when applying MRC models to real-world applications. The experimental results show that MRC models do not perform well on the challenge test set. Moreover, we analyze the behavior of existing models on the challenge test set, which may provide suggestions for future model development. The dataset and codes are publicly available at https://github.com/baidu/DuReader.
Models developed for Machine Reading Comprehension (MRC) are asked to predict an answer from a question and its related context. However, there exist cases that can be correctly answered by an MRC model using BERT, where only the context is provided without including the question. In this paper, these types of examples are referred to as “easy to answer”, while others are as “hard to answer”, i.e., unanswerable by an MRC model using BERT without being provided the question. Based on classifying examples as answerable or unanswerable by BERT without the given question, we propose a method based on BERT that splits the training examples from the MRC dataset SQuAD1.1 into those that are “easy to answer” or “hard to answer”. Experimental evaluation from a comparison of two models, one trained only with “easy to answer” examples and the other with “hard to answer” examples demonstrates that the latter outperforms the former.
In the field of factoid question answering (QA), it is known that the state-of-the-art technology has achieved an accuracy comparable to that of humans in a certain benchmark challenge. On the other hand, in the area of non-factoid QA, there is still a limited number of datasets for training QA models, i.e., machine comprehension models. Considering such a situation within the field of the non-factoid QA, this paper aims to develop a dataset for training Japanese how-to tip QA models. This paper applies one of the state-of-the-art machine comprehension models to the Japanese how-to tip QA dataset. The trained how-to tip QA model is also compared with a factoid QA model trained with a Japanese factoid QA dataset. Evaluation results revealed that the how-to tip machine comprehension performance was almost comparative with that of the factoid machine comprehension even with the training data size reduced to around 4% of the factoid machine comprehension. Thus, the how-to tip machine comprehension task requires much less training data compared with the factoid machine comprehension task.
In this paper, we introduce a simple system Baidu submitted for MRQA (Machine Reading for Question Answering) 2019 Shared Task that focused on generalization of machine reading comprehension (MRC) models. Our system is built on a framework of pretraining and fine-tuning, namely D-NET. The techniques of pre-trained language models and multi-task learning are explored to improve the generalization of MRC models and we conduct experiments to examine the effectiveness of these strategies. Our system is ranked at top 1 of all the participants in terms of averaged F1 score. Our codes and models will be released at PaddleNLP.