Document-based question

Abstract: Document-based Question Answering (DBQA) in Natural Language Processing (NLP) is important but difficult because of the long document and the complex question. Most of previous deep learning methods mainly focus on the similarity computation between two sentences. However, DBQA stems from the reading comprehension in some degree, which is originally used to train and test people’s ability of reading and logical thinking. Inspired by the strategy of doing reading comprehension tests, we propose a unified model based on the human-like reading strategy. The unified model contains three major encoding layers that are consistent to different steps of the reading strategy, including the basic encoder, combined encoder and hierarchical encoder. We conduct extensive experiments on both the English WikiQA dataset and the Chinese dataset, and the experimental results show that our unified model is effective and yields state-of-the-art results on WikiQA dataset.

Abstract: WITH OVER 170,000 STUDENTS taking the advanced placement examination in United States History and 55,000 more taking the European History exam, the Document Based Question, or DBQ, touches the lives of a quarter of a million young people each year Since the introduction of the DBQ in 1975, the essential aim has remained constant: to encourage students to analyze historical evidence and present a coherent interpretive narrative based on that evidence The European History DBQ provides about a dozen documents, while the US History DBQ has a few less, but slightly longer, documents The United States History Test Development Committee asks candidates to provide outside information in their essays, while the European History Committee does not In addition, the US History Test Development Committee announces the half-century from which the DBQ will be drawn a year in advance, while the European History Committee does not Most documents on both exams are passages from letters, speeches, diaries, newspaper accounts and so on There is usually at least one visual document or set of statistics Examples of visual documents are cartoons, photographs, maps and works of art There is usually a single paragraph of historical background There is also, of course, a question Typically, the question asks the student to "analyze and discuss" or "identify and analyze" certain aspects of the situation presented in the documents The student, therefore, must, in

Abstract: Document-based Question Answering aims to compute the similarity or relevance between two texts: question and answer. It is a typical and core task and considered as a touchstone of natural language understanding. In this article, we present a convolutional neural network based architecture to learn feature representations of each question-answer pair and compute its match score. By taking the interaction and attention between question and answer into consideration, as well as word overlap indices, the empirical study on Chinese Open-Domain Question Answering (DBQA) Task (document-based) demonstrates the efficacy of the proposed model, which achieves the best result on NLPCC-ICCPOL 2016 Shared Task on DBQA.