Soft-constrained inference for Named Entity Recognition

Much of the valuable information in supporting decision making processes originates in text-based documents. Although these documents can be effectively searched and ranked by modern search engines, actionable knowledge need to be extracted and transformed in a structured form before being used in a decision process. In this paper we describe how the discovery of semantic information embedded in natural language documents can be viewed as an optimization problem aimed at assigning a sequence of labels (hidden states) to a set of interdependent variables (textual tokens). Dependencies among variables are efficiently modeled through Conditional Random Fields, an indirected graphical model able to represent the distribution of labels given a set of observations. The Markov property of these models prevent them to take into account long-range dependencies among variables, which are indeed relevant in Natural Language Processing. In order to overcome this limitation we propose an inference method based on Integer Programming formulation of the problem, where long distance dependencies are included through non-deterministic soft constraints.     

中小板股票交互相关性研究

以中小板和沪深主板的5 min收益率为研究对象,运用随机矩阵理论的概念和方法对中小板股票进行交互相关性分析,并与主板市场作比较.发现中小板股票问的交互相关性弱于沪深主板股票,当相关系数的均值达到顶峰时,股指刚好分别下跌至各时段的低点,此现象表明中小板股票在下跌过程中有较强的同步性.当股指上升时,不同板块的股票的相关程度...     

Interpolative coding of integer sequences supporting log-time random access

Sequences of integers are common data types, occurring either as primary data or ancillary structures. The sizes of sequences can be large, making compression an interesting option. Effective compression presupposes variable-length coding, which destroys the regular alignment of values. Yet it would often be desirable to access only a small subset of the entries, either by position (ordinal number) or by content (element value), without having to decode most of the sequence from the start. Here such a random access technique for compressed integers is described, with the special feature that no auxiliary index is needed. The solution applies a method called interpolative coding, which is one of the most efficient non-statistical codes for integers. Indexing is avoided by address calculation guaranteeing sufficient space for codes even in the worst case. The additional redundancy, compared to regular interpolative coding, is only about 1 bit per source integer for uniform distribution. The time complexity of random access is logarithmic with respect to the source size for both position-based and content-based retrieval. According to experiments, random access is faster than full decoding when the number of accessed integers is not more than approximately 0.75 · n/log₂n for sequence length n. The tests also confirm that the method is quite competitive with other approaches to random access coding, suggested in the literature.     

在线考试系统设计与实现——以市政设计院考试系统为例

针对目前中国市政工程中南设计研究总院（以下简称中南市政院）考试模式存在的问题,提出并开发了一套适合在设计院内联网使用的在线考试系统。着重介绍了该系统的设计、开发过程及系统实现关键技术。     

基于改进的LBP特征和随机森林相结合的人脸关键点检测方法研究

提出了一种人脸关键点检测方法,该方法用了少量的正面图像,不用归一化人脸图像,而传统的人脸关键点检测方法需要对图像进行严格预处理。随机森林是一种分类器融合算法,可以很好地解决多类分类问题,虽然LBP特征简单,但其可以包含大量的纹理信息。利用改进的LBP特征与随机森林相结合,构成一种对人脸关键点检测的方法。通过高斯平滑图像的LBP特征的提取,对每个点生成特征,计算出有用的特征作为正例,并且与反例集合变为训练集。通过随机森林分类器进行分类,误差率较低,仅在10％左右。     

国内外煤炭价格的非线性特征研究

煤炭是全球基础性能源和战略资源,特别是在以煤炭为主要能源消费的我国,煤炭价格分析是我国经济发展研究中至关重要的问题之一.本文利用BDS 法和替代数据法对我国两种煤炭价格(秦皇岛大同优混(Q5800K)和山西优混(Q5500K))和国外煤炭价格(澳大利亚BJ动力煤现货价)进行了深入研究.研究表明,3种煤炭价格收益率序列不服从正态分布,存在长相关特征并具有非线性结构.国外煤炭价格的非线性特征较国内煤炭价格非线性要强,表明国际煤炭市场的影响因素比国内煤炭市场更加复杂,市场化程度更高.在替代数据法检验时发现在嵌入维数大于5时,上述价格原始数据和替代数据均有显著性差异,表明影响国内外煤炭价格的主要因素至少5个,说明可以建立5维非线性模型进行短期预测,为我国煤炭政策制定及煤炭行业科学发展提供理论指导.     

利用概率的方法证明数学中的一类问题

通过举例说明如何利用概率思想证明代数恒等式和不等式以及求无穷级数的和及极限的方法,从中可以看到概率论与其它学科的联系以及用概率思想解题的美妙之处.进一步阐明概率方法应用的广泛性.     

基于系统抽样技术的随机减量法

将系统抽样方法应用于随机减量法中随机振动信号的提取,对当测量得到的结构响应信号数量很大时如何能较准确地识别结构系统的周期和阻尼比问题进行了研究,给出了结构系统周期和阻尼比估计量的表达式．数值算例结果表明,当结构系统响应信号长度很大时,该方法可较好地解决信号长度与识别精度的矛盾．     

公共图书馆科技期刊服务工作调研——西安图书馆读者科技期刊阅览情况分析报告

本文对西安图书馆科技期刊阅览室读者类型、读者在科技期刊方面的阅读取向以及电子科技期刊利用三个方面做了大量的调查统计工作，揭示了公共图书馆科技期刊阅览室读者服务工作的状况，在此基础上，提出与读者需求匹配的科技期刊服务工作策略。     

关于随机变量的教与学

从随机变量的概念、作用和学习方法等几个方面对随机变量进行了探讨,希望能为概率统计课程的初学者以教师教学提供一些有价值的参考。