Structure-sensitive semantic matching for aggregate question answering over knowledge base

Aggregate question answering essentially returns answers for given questions by obtaining query graphs with unique dependencies between values and corresponding objects. Word order dependency, as the key to uniquely identify dependency of the query graph, reflects the dependencies between the words in the question. However, due to the semantic gap caused by the expression difference between questions encoded with word vectors and query graphs represented with logical formal elements, it is not trivial to match the correct query graph for the question. Most existing approaches design more expressive query graphs for complex questions and rank them just by directly calculating their similarities, ignoring the semantic gap between them. In this paper, we propose a novel Structure-sensitive Semantic Matching(SSM) approach that learns aligned representations of dependencies in questions and query graphs to eliminate their gap. First, we propose a cross-structure matching module to bridge the gap between two modalities(i.e., textual question and query graph). Then, we propose an entropy-based gated AQG filter to remove the structural noise caused by the uncertainty of dependencies. Finally, we present a two-channel query graph representation that fuses the semantics of abstract structure and grounding content of the query graph explicitly. Experimental results show that SSM could learn aligned representations of questions and query graphs to eliminate the gaps between their dependencies, and improves up to 12% (F1 score) on aggregation questions of two benchmark datasets.     

Generating Questions Based on Semi-Automated and End-to-End Neural Network

With the emergence of large-scale knowledge base, how to use triple information to generate natural questions is a key technology in question answering systems. The traditional way of generating questions require a lot of manual intervention and produce lots of noise. To solve these problems, we propose a joint model based on semi-automated model and End-to-End neural network to automatically generate questions. The semi-automated model can generate question templates and real questions combining the knowledge base and center graph. The End-to-End neural network directly sends the knowledge base and real questions to BiLSTM network. Meanwhile, the attention mechanism is utilized in the decoding layer, which makes the triples and generated questions more relevant. Finally, the experimental results on SimpleQuestions demonstrate the effectiveness of the proposed approach.     

Coupling in situ experiments and modeling – Opportunities for data fusion,machine learning,and discovery of emergent behavior

This paper reviews recent studies, that not only includes both experiments and modeling components, but celebrates a close coupling between these techniques, in order to provide insights into the plasticity and failure of polycrystalline metals. Examples are provided of studies across multiple-scales, including, but not limited to, density functional theory combined with atom probe tomography, molecular dynamics combined with in situ transmission electron miscopy, discrete dislocation dynamics combined with nanopillars experiments, crystal plasticity combined with digital image correlation, and crystal plasticity combined with in situ high energy X-ray diffraction. The close synergy between in situ experiments and modeling provides new opportunities for model calibration, verification, and validation, by providing direct means of comparison, thus removing aspects of epistemic uncertainty in the approach. Further, data fusion between in situ experimental and model-based data, along with data driven approaches, provides a paradigm shift for determining the emergent behavior of deformation and failure, which is the foundation that underpins the mechanical behavior of polycrystalline materials.     

有水气藏开发早期可采储量的评价方法

有水气藏是天然气藏中的一个基本气藏类型，其采收率受多种因素的影响，如地面集输压力、井筒混合气水流动的液柱压力和储层物性等，目前普遍采用经验取值、物质平衡和气藏数值模拟等方法进行研究，但由于受开发评价阶段试采资料偏少和气藏数值模拟无法模拟井筒流动等因素的影响，早期采收率计算往往偏差较大，针对这些问题，从气田开发一体化的特点出发，提出了一种全面考虑地层→井筒→地面流动条件下，有水气藏（含凝析有水气藏）发废弃压力的计算方法，发展了利用物质平衡早期预测有水气藏采收率的理论，方法可靠、适用性强，能够推广应用。     

微机控制托辊管筒加工机床的设计

介绍了一种生产效率高、加工质量稳定、利用单片机控制液压传动系列带式输送机托辊管筒自动加工机床的设计过程。     

计算机基础课题库系统中试卷生成模型研究

探讨了计算机基础课程题库系统中题型难度系数确定模型、试题在章节中的分布模型、各级难度题型分布模型，认为利用这些模型和随机函数能够确定计算机基础课程相应章节中相应等级难度的试题中抽取相应数量的试题，再利用数据库的检索功能实现试卷的自动生成。     

隧道工程施工中应注意的问题

通过对海南省东线高速公路大茅山隧道施工的全程监理,阐明了在施工中应重点注意的几个问题,在此基础上,总结出隧道工程质量监督中必须做到的几点,以确保隧道工程施工安全及工程质量.     

The summary of researches on ADS in China

The conceptual study on the accelerator driven system (ADS) that lasted for about five years ended in 1999 in China. As one project of the National Basic Research Program of China (973 Program) in the energy domain, which is sponsored by the China Ministry of Science and Technology (MOST), a five-year-program of fundamental research of ADS physics and related technology was launched in 2000 and passed national review at the end of 2005. The China Institute of Atomic Energy (CIAE), the Institute of High Energy Physics (IHEP), the Institute of Heavy Ion Physics in Peking University (PKU-IHIP) and other institutions jointly carried out the research. The research activities were focused on HPPA physics and technology, reactor physics of external source driven sub-critical assembly, nuclear data base and material study. For HPPA, a high current injector consisting of an ECR ion source, LEBT and an RFQ accelerating structure of 3.5 MeV was built. In reactor physics study, a series of neutron multiplication experimental study has been carried out and still being done. The VENUS facility has been constructed as the basic experimental platform for neutronics study in ADS blanket. It’s a zero power sub-critical neutron multiplying assembly driven by external neutron produced by a pulsed neutron generator. The theoretical, experimental and simulation study on nuclear data, material properties and nuclear fuel circulation related to ADS is carrying on to provide the database for ADS system analysis. The main results on ADS related researches will be reported.     

基于进化计算的兴趣建模

为实现对个体的兴趣建模,提出基于进化计算的建模(Evolutionary Computation Based Modeling,ECBM)方法。该方法采用定性和定量相结合的方式对个体的兴趣系统进行建模,首先通过定性分析构建个体的兴趣模型框架,然后通过进化计算(Evolutionary Computation)方法对模型参数自动进化。实验结果表明,ECBM能够通过数据驱动模式优化模型参数,并利用更新信息进化模型,使其不断逼近真实系统。     

模型驱动下的Web应用系统自动生成

为了提高Web应用系统开发效率,提出了模型驱动下的Web应用系统自动生成方法。这种生成方法是以Me—taEdit＋作为元建模工具,首先创建Web应用系统元模型、定制DSL,进而建立Web应用系统领域模型,然后通过MetaEdit＋提供的生成器定义语言MERL,软件开发人员可以很方便地设计出Web应用系统所需的JSP生成器、Servlet生成器、Jayabeans生成器和数据库生成器,从Web应用系统图形模型直接生成整个Web应用系统。最后通过一个WebShopping实例详细介绍了模型驱动下的Web应用系统生成方法及生成过程。经测试,所生成的Web应用系统可以在Windows操作系统中的Web应用服务器上正确运行。