第四代语言NH4GL系统

本文介绍了第四代语言系统NH4GL,并论述了其设计准则和实现方案。     

基于主题模型的结构化Web服务发现机制

提出了基于主题模型(topic model)的结构化Web服务发现机制.利用LDA(Latent Dirichlet Allocation)生成概率模型,将Web服务(Web service)建模为结构化文本文档.一个文档视作主题的概率分布,主题又由关键词的概率分布组成,从而提供基于主题的Web服务检索.同时,利用Web服务的结构化特性,将Web服务描述文档表示为有向无环图,利用n阶谱核测量Web服务文档的相似度,实现Web服务结构化信息的发现.通过实验分析对比,基于主题模型的结构化Web服务发现机制有效提高了Web服务发现的效率和精确率.      

基于边缘检测与颜色目标定位的图像检索

提出一种先利用图像边缘信息缩小检索范围。再利用图像颜色信息进行目标定位的图像检索算法。首先。提取图像边缘信息。利用图像位错率进行边缘信息比较,位错率越小图像边缘信息越接近;然后,利用主色调确定图像的目标块,进行目标子块间的比较。算法忽略图像中不重要的背景因素,有效利用重要的目标因素。从而提高了检索效率。仿真实验证明该算法比单一使用边缘检测、颜色目标定位具有更好的检索性能。     

面向中文问答的信息检索系统及评测

为了解决中文问答系统中检索有效性的问题,提出并构建了一个面向中文问答的信息检索系统,在新进展的基础上,该系统通过利用问答系统中问题分类信息,对查询进行优化处理,从而改进问答系统中检索结果的质量,在独立的检索评测中正确率达到0.5013,在问答系统的整体评测中F3值达到0.2231,取得了较好的效果.     

基于粒子群算法的绳系卫星展开与回收控制

绳系卫星的展开与回收是绳系卫星应用的基础问题,但通过改变系绳拉力来实现展开和回收的方法尚未得到充分研究.将系绳拉力作为控制力,并在时间域内离散,然后利用粒子群优化算法在整个可行域中寻找满足优化指标的最优值,从而得到控制力的变化规律.该控制方法可以使绳系卫星系统在最小摆动、最短时间等最优指标意义下,迅速展开到平衡位置,同时也适用于绳系卫星的回收.仿真结果表明：该控制方法同基于控制展开与回收速度的控制方法相比,展开和回收的时间短、侧向摆动小,具有一定优越性.     

Synergetic retrieval of terrestrial AOD from MODIS images of twin satellites Terra and Aqua

Aerosol optical depth (AOD) is one of the most important indicators of atmospheric pollution. It can be retrieved from satellite imagery using several established methods, such as the dark dense vegetation method and the deep blue algorithm. All of these methods require estimation of surface reflectance prior to retrieval, and are applicable to a certain pre-designated type of surface cover. Such limitations can be overcome by using a synergetic method of retrieval proposed in this study. This innovative method is based on the fact that the ratio K of surface reflectance at different angles/geometries is independent of wavelength as reported by Flowerdew and Haigh (1995). An atmospheric radiative transfer model was then established and resolved with the assistance of the ratio K obtained from two Moderate Resolution Imaging Spectroradiometer (MODIS) spectral bands acquired from the twin satellites of Terra and Aqua whose overpass is separated by three hours. This synergetic method of retrieval was tested with 20 pairs of MODIS images. The retrieved AOD was validated against the ground observed AOD at the Taihu station of the AErosol RObotic NETwork (AERONET). It is found that they are correlated with the observations at a coefficient of 0.828 at 0.47 μm and 0.921 at 0.66 μm wavelengths. The retrieved AOD has a mean relative error of 25.47% at 0.47 μm and 24.3% at 0.66 μm. Of the 20 samples, 15 and 17 fall within two standard error of the line based observed AOD data on the ground at the 0.47 μm and 0.66 μm, respectively. These results indicate that this synergetic method can be used to reliably retrieve AOD from the twin satellites MODIS images, namely Terra and Aqua. It is not necessary to determine surface reflectance first.     

专利信息检索技术

专利信息检索作为快速获取专利信息的主要手段,在专利信息应用过程中发挥重要作用。本文在分析当前专利信息检索涉及的关键技术的基础上,进一步阐述了基于概念的专利信息检索技术。本文以对专利文献的内容分析为基础,按照文献中各部分的功能进行组织表示,对不同部分制定相应的特征选择策略。同时,对基于概念的专利信息检索技术中的两个关键问题即查询扩展和专利文献相似度计算方法进行了阐述。     

基于知识图谱的遥感图像检索定位方法

为了快速、准确地在海量遥感图像库中检索出所需的遥感图像资源，提出了一种基于知识图谱的遥感图像检索定位方法。为了充分利用遥感图像中所蕴含的地理语义信息，方法中使用知识图谱节点代表遥感图像，根据图像间的相似程度和地理位置在节点间建立关系。通过知识图谱缩小图像检索空间，加快检索速度，准确地定位遥感图像所示区域的地理信息，并返回最相似的图像。将此方法应用于卫星自主导航领域中，与传统的线性检索相比，图像检索的时间缩短了50%，准确率也得到提高。     

Pose matters: Pose guided graph attention network for person re-identification

Person re-Identification (reID), aiming at retrieving a person across different cameras, has been playing a more and more important role in the construction of smart city and social security. For deep-learning-based reID methods, it has been proved that using local feature together with global feature could help to give robust representation for person retrieval. Human pose information can provide the locations of human skeleton to effectively guide the network to pay more attention to these key areas, and can also help to reduce the noise distractions from background or occlusions. Based on human pose, a Pose Guided Graph Attention (PGGA) network is proposed in this paper, which is a multi-branch architecture consisting of one branch for global feature and two branches for local key-point features. A graph attention convolution layer is carefully designed to re-assign the contribution weight of each extracted local feature by modeling the similarity relations. The experimental results demonstrate the effectiveness of our approach on discriminative feature learning. Our model achieves the state-of-the-art performance on several mainstream evaluation datasets. A plenty of ablation studies and different kinds of comparison experiments are conducted to prove the effectiveness of this work, including the tests on occluded datasets and cross-domain datasets. Moreover, we further design supplementary tests in practical scenario to indicate the advantage of our work in real-word applications.