基于Z规格的LR(k)形式化分析及验证

在编译器的构造中,常由于语义的二义性等问题导致不正确的目标程序.为解决此问题,提出了一种新型的语法及语义正确性验证方案,即建立LR (k)文法和Z规格说明的联系,以此构造LR (k)文法的形式化描述及其形式化验证.实验结果表明,该方案能有效描述并检测LR (k)文法分析器中的语法错误及语义二义性,有助于提高分析器的有效性.     

一个高效的语法分析器生成工具

VPGE(VisualParserGenerationEnvironment)是一个可视化语法分析器集成开发环境,除了具有良好的界面和强大的调试功能,其LALR(1)分析器的生成速度达到并超过公认的分析器生成速度最快的LALR(1)分析器自动生成器Bison,所能处理的文法规模也优于Bison.本文在DeRemer和Pen-nello的LALR(1)分析器自动生成基本原理的基础上,在设计数据结构和算法实现中采用了大量优化技术.     

一个高效的语法分析器生成工具

VPGE(Visual Parser Generation Environment)是一个可视化语法分析器集成开发环境,除了具有良好的界面和强大的调试功能,其LALR(1)分析器的生成速度达到并超过公认的分析器生成速度最快的LALR(1)分析器自动生成器Bison,所能处理的文法规模也优于Bison.本文在DeRemer和Pennello的LALR(1)分析器自动生成基本原理的基础上,在设计数据结构和算法实现中采用了大量优化技术.     

VPGE:一个LALR(1)分析器的可视化生成和断点调试系统

LALR(1)分析程序生成系统在编译器构造领域以外被许多普通软件开发者学习和使用.为帮助用户理解LALR(1)分析器方法,编写出正确、完整、无语法分析冲突的文法规范,严格定义了使用LALR(1)分析器生成器时用户可能遇到的几类文法问题,描述一个为帮助用户解决这些问题而开发的LALR(1)分析器可视化和断点调试系统VPGE.VPGE以多种视图显示LALR(1)分析器的数据结构,包括状态栈、符号栈、输入符号串、分析树和底层的自动机,支持LR分析动作的单步执行和断点调试.性能实验结果表明,VPGE比GNU的Bison有更快的分析器生成速度,从而提供了一个LALR(1)文法及分析器的快速交互式调试环境.     

消除C++分析器分析过程中的二义性

C++语言的文法十分复杂,采用LL(k)或者LR(k)分析方法都面临二义性问题。另外,内联函数分析,名字查找,模板分析,需要采用特殊的方法解决,本文主要阐述了这些问题,并结合实际的分析器JLParser提出了其解决方法。     

案例教学法在《编译原理》课程中的研究

本文将案例法引入了《编译原理》课程的教学,发挥案例法的优势,以二义文法为主线,探讨了二义文法和递归文法、LL（1）文法、LR（0）文法和SLR（1）文法的关系,找出了一些判定文法有无二义性的充分条件,同时研究了解决二义性的方法,加深了学生对二义文法的理解。     

程序设计语言的GLR优化分析

阐述了在程序设计语言语法分析器的构造中采用通用LR(generalized LR,简称GLR)分析算法的动机.提出了一个多层次的优化策略,加快了GLR分析器的分析速度.为基本的GLR算法增加了必要的运行时控制机制,以实现语法分析时调用文法规则附带的语义动作,化解输入串的二义性,同时避免GLR分析器可能存在的语义动作延迟问题.优化后的算法已在一个可视化语法分析器自动生成环境VPGE中实现.实验结果表明,在分析确定性的编程语言时,自动生成的GLR分析器的分析速度与自由软件基金会的Bison生成的LALR(1)分析器的分析速度有可比性.     

LR(k)任意文法位置的断点调试方法

LR(k)文法能描述所有确定型上下文无关语言,广泛应用于各类分析器生成器中.传统的LR(k)文法断点调试方法仅支持在产生式右部末尾设置断点(后文简称尾部断点),不支持在产生式右部中间位置设置断点(后文简称中间断点),这给分析器的开发和调试带来了不便.文中提出了一种新颖的LR(k)文法断点调试方法,不但支持传统的尾部断点,还支持中间断点.该方法可显著增加可利用的断点数量,可以跟踪到更细粒度的文法成分,从而帮助用户更好地进行文法调试,降低分析器的开发难度.     

SLR(1)词/语法分析的自动构造

主要讨论了SLR（1）词／语法分析器的自动构造，自动生成的SLR（1）语法分析器可从自动生成的SLR（1）词法分析器中获得测试数据进行语法分析，论文相关的软件除可用于编译原理课程教学演示外，还可用于实际编译程序的SLR（1）词／语法分析器的自动构造。由于具有SLR（1）分析表自动构造功能，因此该软件还可作为软件工具用于程序设计语言本身的设计和构造。     

编译原理中文法的教学探讨

在分析正规式、正规文法和上下文无关文法的基础上,着重讨论上下文无关文法中涉及到的一些重要的文法,例如二义文法、递归文法、LL（1）文法、LR（0）文法和SLR（1）文法,并以二义文法为主线,分析二义文法和递归文法、LL（1）文法、LR（0）文法和SLR（1）文法的关系,找出一些判定文法有无二义性的充分条件。     

基于稀疏逻辑回归和多元融合算法的慢性肾病进展预测模型

只有一部分慢性肾病（Chronic Kidney Disease， CKD）3期的患者会进展到4期，观察临床数据发现进展和非进展患者部分生理指标有较大的区别。本文首次将基于L1/2范数正则化的逻辑回归（Sparse Logistic Regression， SLR）用于筛选影响CKD患者进展的关键因素，然后利用SLR、支持向量机（SVM）、提升决策树（AdaBoost Decision Tree, BOOSTDT）建立进展风险预测模型。另外，本文引入堆叠算法Stacking（STKSSD）克服样本量不足使得模型泛化性能不稳定的缺陷。作为对比，本文分别利用神经网络（ANN）、循环神经网络（BLSTM）对数据建模。实验结果表明，当SLR算法选择磷、血清肌酐等11个关键特征时， STKSSD融合模型效果最好，其中测试查全率、查准率、F1值分别为86.97%、92.86%和89.82%。     

Risks and risk mitigation in global software development: A tertiary study

ContextThere is extensive interest in global software development (GSD) which has led to a large number of papers reporting on GSD. A number of systematic literature reviews (SLRs) have attempted to aggregate information from individual studies.ObjectiveWe wish to investigate GSD SLR research with a focus on discovering what research has been conducted in the area and to determine if the SLRs furnish appropriate risk and risk mitigation advice to provide guidance to organizations involved with GSD.MethodWe performed a broad automated search to identify GSD SLRs. Data extracted from each study included: (1) authors, their affiliation and publishing venue, (2) SLR quality, (3) research focus, (4) GSD risks, (5) risk mitigation strategies and, (6) for each SLR the number of primary studies reporting each risk and risk mitigation strategy.ResultsWe found a total of 37 papers reporting 24 unique GSD SLR studies. Major GSD topics covered include: (1) organizational environment, (2) project execution, (3) project planning and control and (4) project scope and requirements. We extracted 85 risks and 77 risk mitigation advice items and categorized them under four major headings: outsourcing rationale, software development, human resources, and project management. The largest group of risks was related to project management. GSD outsourcing rationale risks ranked highest in terms of primary study support but in many cases these risks were only identified by a single SLR.ConclusionsThe focus of the GSD SLRs we identified is mapping the research rather than providing evidence-based guidance to industry. Empirical support for the majority of risks identified is moderate to low, both in terms of the number of SLRs identifying the risk, and in the number of primary studies providing empirical support. Risk mitigation advice is also limited, and empirical support for these items is low.     

基于频域相位方差加权的线性调频信号检测方法

针对未知线性调频信号的检测问题,依据线性调频信号相位比较稳定这一特征,提出一种基于频域相位方差加权的线性调频信号检测方法。该方法利用线性调频信号频率单元对应相位比较稳定,背景噪声频率单元对应相位比较随机的特点,对各频率单元进行相位方差加权,可以进一步抑制背景噪声能量干扰,增强线性调频信号检测信噪比增益,实现对未知线性调频信号的检测。仿真条件下,在输入平均谱级比大于-10 dB时,相比相位差分对齐法,该方法所得最终线性调频信号频率单元与噪声频率单元的平均谱级比得到了进一步提高,且随着输入平均谱级比越高,输出线性调频信号频率单元与噪声频率单元的平均谱级比提高越多。理论分析和实验结果表明:该方法可以有效增强信号能量,抑制噪声,提高信噪比。     

Improving sustainability of BitTorrent darknets

Private BitTorrent community, known as “BitTorrent Darknet” or “Private Tracker (PT)”, has received much attention in the research literature recently. Different from public BitTorrent community, PT can only be accessed by its registered members, and it can provide ultra-high downloading speed due to its effective Sharing Ratio Enforcement (SRE) incentive mechanism which stimulates members to contribute content as much as possible. Although the downloading performance in PTs with high Seeder-to-Leecher Ratio (SLR) is much better than in public BitTorrent communities, our measurements show that SRE mechanism can induce “Poor Downloading Motivation” problem for members who want to increase their sharing ratios to survive and thrive in PT. This problem may discourage PT members’ enthusiasm from contributing to community. To improve sustainability of PTs, we adopt the Predator–prey model to analyze high SLR phenomenon, study the optimal stable SLR range to PTs. Moreover, we develop a queuing model to simulate the seeding/leeching process in a single torrent in PT, and try to maximize the swarming performance with minimum seeding peers. This solution can help PT achieve the optimal SLR range.     

Gait parameters as predictors of slip severity in younger and older adults

This study investigated the association between slip severity and pre-slip gait characteristics of younger and older subjects. Sixteen younger and eleven older healthy adults walked onto an unexpectedly slippery surface. Slip severity was categorized as either hazardous or non-hazardous using a 1.0 ms peak slip velocity threshold. The results showed that hazardous slips were associated with greater step lengths (normalized by leg length) (SLR), larger and more rapidly changing foot - floor angles (FFA) at heel strike, and increased cadence across the two subject groups. Older subjects were found to walk with shorter SLR and with smaller and more slowly changing FFA at heel strike compared to younger subjects. However, both younger and older subjects experienced hazardous slips at the same rate. A logistic regression model relating SLR and cadence to slip severity predicted that increased SLR and decreased cadence would result in increased probability of hazardous slip (R2 = 0.45, chi2 = 15.30, p<0.01). A second logistic regression model relating FFA with slip severity predicted that increased FFA would result in increased probability of hazardous slip (R2 = 0.53, chi2 = 16.55, p<0.01). These results suggest that gait characteristics prior to foot contact play an important role in the severity of an ensuing slip. The finding that older adults experienced hazardous slips at the same rate as young adults even though their SLR and FFA are smaller suggests that age is also playing a role in other aspects of postural control that impact slip severity.     

Computational Reconstruction of Cognitive Music Theory

In order to obtain a computer-tractable model of music, we first discuss what conditions the music theory should satisfy from the various viewpoints of artificial intelligence and/or other computational notions. Then, we look back on the history of cognitive theory of music, i.e., various attempts to represent our mental understandings and to show music structures. Among which, we especially pay attention to the Generative Theory of Tonal Music (GTTM) by Lehrdahl and Jackendoff, as the most promising candidate of cognitive/computational theory of music. We briefly overview the theory as well as its inherent problems, including the ambiguity of its preference rules. By our recent efforts, we have solved this ambiguity problem by assigning parametrized weights, and thus we could implement an automatic tree analyzer. After we introduce the system architecture, we show our application systems.     

Large-Vocabulary Continuous Sign Language Recognition Based on Transition-Movement Models

The major challenges that sign language recognition (SLR) now faces are developing methods that solve large-vocabulary continuous sign problems. In this paper, transition-movement models (TMMs) are proposed to handle transition parts between two adjacent signs in large-vocabulary continuous SLR. For tackling mass transition movements arisen from a large vocabulary size, a temporal clustering algorithm improved from k-means by using dynamic time warping as its distance measure is proposed to dynamically cluster them; then, an iterative segmentation algorithm for automatically segmenting transition parts from continuous sentences and training these TMMs through a bootstrap process is presented. The clustered TMMs due to their excellent generalization are very suitable for large-vocabulary continuous SLR. Lastly, TMMs together with sign models are viewed as candidates of the Viterbi search algorithm for recognizing continuous sign language. Experiments demonstrate that continuous SLR based on TMMs has good performance over a large vocabulary of 5113 Chinese signs and obtains an average accuracy of 91.9%     

面向中等词汇量的中国手语视觉识别系统

手语识别的研究和实现具有重要的学术价值和广泛的应用前景．提出了基于混合元捆绑的隐马尔可夫模型（TMHMM）用于视觉手语识别．TMHMM的模型刻画精度接近于连续隐马尔可夫模型,因此能保证最终的识别率不会明显降低,同时通过混合元捆绑降低计算成本,有效地提高识别速度．在特征提取方面,提出的层次型特征描述方案更加适合于中等或更大词汇量的手语识别．在此基础上。通过集成鲁棒的双手检测、背景去除和瞳孔检测等技术,实现了一个面向中等词汇量的中国手语视觉识别系统．实验结果表明,提出的方法能较好地实现常规背景中的中等词汇量的手语识别．