下载文档原格式
《第二语言习得研究》全书概要第一章绪论第一节第二语言习得的一般概念第二语言习得(second language acquisition)，简称SLA，是指人们在获得母语(第一语言)的基础上习得另一种或几种语言的过程。
也叫“二语习得”。
一、母语与目的语母语指学习者所属种族、社团使用的语言，也称作“本族语”。
目的语，也叫目标语，一般指学习者正在学习的语言。
二、第一语言与第二语言第一语言指儿童幼年最先接触和习得的语言，在此后习得的语言就是第二语言。
三、习得与学习克拉申（Krashen）认为成人L2学习者有两种独立的语言获得方式，两者在获得方式、心理过程、所获得的知识类型、作用等方面都不同Krashen认为，习得的知识与学得的知识是相互独立的两种知识。
学得的知识无法转换成习得的知识。
即所谓“无接口观点”(non-interface position)。
自然习得研究的证据表明，在习得情况下并未发生学习的过程，有时候，学习者可以先学会某个规则，但是并没有习得这个规则。
也就是说学习过程并不一定导致习得过程的发生。
无接口观点的证据：Seliger（1979）的证据：他让在课堂上让学习者描述一些图片，然后分析这些学习者使用冠词a与an的情况；此外，他让这些学习者陈述关于冠词用法的相关规则。
调查分析表明，学习者的实际语言表达与其元语言知识不相关。
也就是说，“习得”与“学习”的确是彼此独立的。
第二语言习得是指学习者在目的语国家学习目的语。
外语习得指学习者所学的语言在本国不是作为整个社团的交际工具，而且学习者所学的语言主要是在课堂上学习的。
五、自然的第二语言习得与有指导的第二语言习得：前者指在自然的社会环境下以交际的方式获得第二语言；后者指在课堂教学环境中以教学指导的方式获得第二语言。
语言能力是一种反映交际双方语言知识的心理语法。
语言表达指交际双方在语言的理解与生成过程中对其内在语法的运用。
一、第二语言习得研究与语言学：语言学是一个古老的学科，第二语言习得研究则是一个年轻的学科。
语言学考研真题和答案第一章语言学Fill in the blanks1. Human language is arbitrary. This refers to the fact that there is no logical or intrinsic connection between a particular sound and the _______it is associated with. （人大2007研）meaning 语言有任意性，其所指与形式没有逻辑或内在联系2. Human languages enable their users to symbolize objects, events and concepts which are not present (in time and space) at the moment of communication. This quality is labeled as _______. （北二外2003研）displacement 移位性指人类语言可以让使用者在交际时用语言符号代表时间和空间上不可及的物体、事件和观点3. By duality is meant the property of having two levels of structures, such that units of the _______ level are composed of elements of the __________ level and each of the two levels has its own principles of organization. （北二外2006研）primary, secondary 双重性指拥有两层结构的这种属性，底层结构是上层结构的组成成分，每层都有自身的组合规则4. The features that define our human languages can be called _______ features. （北二外2006）design人类语言区别于其他动物交流系统的特点是语言的区别特征，是人类语言特有的特征。
2001 Passage 1 1. No clear－cut distinction can be drawn between professionals and amateurs in science: exceptions can be found to any rule. Nevertheless,the word “amateur” does carry a connotation that the person concerned is not fully integrated into the scientific community and,in particular,may not fully share its values. 「译⽂」在科学领域内，专业与业余之间没有明显的区分：任何规律都有其例外。
但是“业余”这个词的确具有某种（特殊的）含义，即相关⼈员并没有完全融⼊某个科学群体，尤其是指他们也许并不完全认同这个群体的价值观。
「析句」第⼀个句⼦实际上是因果复合句，主句是No clear cut… in science，后⾯是⼀个省略了as的从句，以说明原因。
在第⼆个句⼦中,nevertheless是转折词，表⽰上下⽂语意的转折，其主句是the word “amateur” does carry a connotation，后⾯跟了⼀个同位语从句，其中有两个并列的谓语成分。
「讲词」 clear-cut意为“明显的，明确的，清楚的”。
This is a clear-cut victory for our team.（对于我们的球队来说，这是⼀场完胜。
） distinction意为“区别，差别，特性，声望”。
在表⽰“区别，差别”时，它与 difference⽐更强调区别或辨别的过程。
He is a diplomat of distinction.（他是⼀位卓越的外交官。
） 2. The trend was naturally most obvious in those areas of science based especially on a mathematical or laboratorytrai ning,and can be illustrated in terms of the development of geology in the United Kingdom. 「译⽂」特别是在以数学和实验室训练为基础的科学领域，这种倾向⾃然最为明显，它可以通过英国的地质学发展过程来加以说明。
英语三级试题及答案一、听力理解（共20分）A. 短对话理解（每题1分，共5分）1. What is the man going to do?A) Buy a book. B) Return a book. C) Borrow a book.Answer: B2. Where does the conversation most probably take place?A) In a library. B) In a bookstore. C) In a classroom.Answer: A3. What does the woman mean?A) She is busy. B) She is interested in the book. C) Sheis going to the library.Answer: C4. Why does the man suggest going to the library?A) To find a better book. B) To study in a quiet place. C) To meet a friend.Answer: B5. What is the relationship between the two speakers?A) Friends. B) Classmates. C) Teacher and student.Answer: AB. 长对话理解（每题2分，共10分）听下面一段较长的对话，回答6至10题。
6. What is the main topic of the conversation?A) A birthday party. B) A school project. C) A travel plan.Answer: A7. When is the party going to be held?A) Next Friday. B) This weekend. C) Tomorrow.Answer: B8. What does the woman suggest the man do?A) Buy a gift. B) Prepare a speech. C) Make a reservation. Answer: A9. What does the man think of the woman's idea?A) He agrees with her. B) He disagrees with her. C) Hehas no opinion.Answer: A10. What will the woman probably do next?A) Call her friend. B) Go shopping. C) Make a list.Answer: CC. 短文理解（每题2分，共5分）听下面一段短文，回答11至15题。
高一人工智能英语阅读理解25题1<背景文章>Artificial intelligence (AI) has become one of the most talked - about topics in recent years. AI can be defined as the simulation of human intelligence processes by machines, especially computer systems. These processes include learning, reasoning, problem - solving, perception, and language understanding.The development of AI has a long history. It started in the 1950s when the concept was first introduced. In the early days, AI research focused on simple tasks like playing games and solving basic mathematical problems. However, with the development of computer technology and the increase in data availability, AI has made great strides.AI has found applications in various fields. In the medical field, AI can assist doctors in diagnosing diseases. For example, it can analyze medical images such as X - rays and MRIs to detect early signs of diseases that might be missed by human eyes. In education, AI - powered tutoring systems can provide personalized learning experiences for students. They can adapt to the individual learning pace and style of each student, helping them to better understand difficult concepts. In the transportation industry, self - driving cars, which are a significant application of AI, are expectedto revolutionize the way we travel. They can potentially reduce traffic accidents caused by human error and improve traffic efficiency.However, AI also brings some potential negative impacts. One concern is the impact on employment. As AI systems can perform many tasks that were previously done by humans, there is a fear that many jobs will be lost. For example, jobs in manufacturing, customer service, and some administrative tasks may be at risk. Another issue is the ethical considerations. For instance, how should AI make decisions in life - or - death situations? And there are also concerns about data privacy as AI systems rely on large amounts of data.1. <问题1>What is the main idea of this passage?A. To introduce the development of computer technology.B. To discuss the applications and impacts of artificial intelligence.C. To explain how to solve problems in different fields.D. To show the importance of data in AI systems.答案：B。
*国家自然科学基金资助项目(N o .60875012,60905005)收稿日期:2009-12-21;修回日期:2010-01-27作者简介 高隽,男,1963年生,教授,博士生导师,主要研究方向为图像理解、智能信息处理、光电信息处理等.E-m a i:l gao j un @hfut .edu .cn .谢昭,男,1980年生,博士,讲师,主要研究方向为计算机视觉、智能信息处理、模式识别.张骏,女,1984年生,博士研究生,主要研究方向为图像理解、认知视觉、机器学习.吴克伟,男,1984年生,博士研究生,主要研究方向为图像理解、人工智能.图像语义分析与理解综述*高 隽 谢 昭 张 骏 吴克伟(合肥工业大学计算机与信息学院合肥 230009)摘 要 语义分析是图像理解中高层认知的重点和难点,存在图像文本之间的语义鸿沟和文本描述多义性两大关键问题.以图像本体的语义化为核心,在归纳图像语义特征及上下文表示的基础上,全面阐述生成法、判别法和句法描述法3种图像语义处理策略.总结语义词汇的客观基准和评价方法.最后指出图像语义理解的发展方向.关键词 图像理解,语义鸿沟,语义一致性,语义评价中图法分类号 T P 391.4I m age Se m antic Anal ysis and Understandi ng :A R eviewGAO Jun ,XI E Zhao ,Z HANG Jun ,WU Ke -W ei(S chool of C o m puter and Infor m ation,H e fei University o f T echnology,H efei 230009)ABSTRACTSe m antic ana l y sis is the i m portance and diffi c u lty of high -level i n terpretati o n i n i m age understandi n g ,i n wh ich there are t w o key issues of tex-t i m age se m an tic gap and tex t descri p ti o n po lyse m y .Concentrating on se m antizati o n o f i m ages onto logy ,three soph i s tica ted m et h odolog ies are round l y rev ie w ed as generati v e ,d iscri m ina ti v e and descriptive gra mm ar on the basis of conc l u d i n g i m ages se m antic fea t u res and context expression .The ob jective benchm ark and eva l u ation for se m an tic vocabu lary are i n duced as w e l.l F i n ally ,the summ arized directions fo r furt h er researches on se m antics i n i m age understand i n g are discussed i n tensively .K ey W ords I m age Understanding ,Se m antic G ap ,Se m an tic Consistency ,Se m an tic Evalua ti o n1 引 言图像理解(I m age Understandi n g ,I U )就是对图像的语义解释.它是以图像为对象,知识为核心,研究图像中何位置有何目标(what is w here)、目标场景之间的相互关系、图像是何场景以及如何应用场景的一门科学.图像理解输入的是数据,输出的是知识,属于图像研究领域的高层内容[1-3].语义(Se -第23卷 第2期 模式识别与人工智能 V o.l 23 N o .2 2010年4月 PR &A I A pr 2010m antics)作为知识信息的基本描述载体,能将完整的图像内容转换成可直观理解的类文本语言表达,在图像理解中起着至关重要的作用.图像理解中的语义分析在应用领域的潜力是巨大的.图像中丰富的语义知识可提供较精确的图像搜索引擎(Searching Eng i n e),生成智能的数字图像相册和虚拟世界中的视觉场景描述.同时,在图像理解本体的研究中,可有效形成/数据-知识0的相互驱动体系,包含有意义的上下文(Context)信息和层状结构(H ierarchica-l S truct u red)信息,能更快速、更准确地识别和检测出场景中的特定目标(如,识别出场景中的/显示器0,根据场景语义知识可自动识别附近的/键盘0).尽管语义分析在图像理解中处于非常重要的位置,但传统的图像分析方法基本上全部回避了语义问题,仅针对纯粹的图像数据进行分析.究其原因主要集中于两方面:1)图像的视觉表达和语义之间很难建立合理关联,描述实体间产生巨大的语义鸿沟(Se m antic Gap);2)语义本身具有表达的多义性和不确定性(Am bigu ity).目前,越来越多的研究已开始关注上述/瓶颈0,并致力于有效模型和方法以实现图像理解中的语义表达.解决图像理解中的语义鸿沟需要建立图像和文本之间的对应关系,解决的思路可大致分为三类.第一条思路侧重于图像本身的研究,通过构建和图像内容相一致的模型或方法,将语义隐式地(I m p lici-t l y)融入其中,建立/文本y图像0的有向联系,核心在于如何将语义融于模型和方法中.采用此策略形成的研究成果多集中于生成(Generati v e)方式和判别(D iscri m inati v e)方式中.第二条思路从语义本身的句法(G ra mm ar)表达和结构关系入手,分析其组成及相互关系,通过建立与之类似的图像视觉元素结构表达,将语义描述和分析方法显式地(Exp lici-t l y)植入包含句法关系的视觉图中,建立/图像y文本0的有向联系.核心在于如何构建符合语义规则的视觉关系图.第三条思路面向应用,以基于内容的图像检索(I m age Retrieval)为核心,增加语义词汇规模,构建多语义多用户多进程的图像检索查询系统.解决语义本身的多义性问题需要建立合理的描述规范和结构体系.Princeton大学的认知学者和语言学家早在20世纪80年代就研究构建了较合理统一的类树状结构.如今已被视为视觉图像研究领域公认的语义关系参考标准,用于大规模图像数据集的设计和标记中,有效归类统一了多义性词语.此外,一些客观的语义检索评价标准也在积极的探索过程中.本文将对上述两个图像语义理解中的问题进行方法提炼和总结.针对语义鸿沟问题,介绍已有模型和方法的处理策略.还采用较完备的图像语义/标尺0(B ench m ark)解决语义的主观多义性.2图像内容的语义分析图像内容描述具有/像素-区域-目标-场景0的层次包含关系,而语义描述的本质就是采用合理的构词方式进行词汇编码(Encodi n g)和注解(Annota-tion)的过程.这种过程与图像内容的各层描述密切相关,图像像素和区域信息源于中低层数据驱动,根据结构型数据的相似特性对像素(区域)进行/标记0(Labeli n g),可为高层语义编码提供有效的低层实体对应关系.目标和场景的中层/分类0(C ategor-i zati o n)特性也具有明显的编码特性,每一类别均可视为简单的语义描述,为多语义分析的拓展提供较好的原型描述.本节将针对前述的语义鸿沟问题介绍常用的图像语义表示方法和分析策略.2.1语义化的图像特征图像内容的语义分析借鉴文本分析策略.首先需要构建与之相对应的对象,整幅图像(I m age)对应整篇文档(Docum ent),而文档中的词汇(Lex icon)也需要对应相应的视觉词汇(V isua lW ord).视觉词汇的获取一般通过对图像信息的显著性分析提取图像的低层特征,低层特征大多从图像数据获取,包括简单的点线面特征和一些特殊的复杂特征,再由鲁棒的特征表达方式生成合适的视觉词汇,视觉词汇一般具有高重用性和若干不变特性.点特征提取以图像中周围灰度变化剧烈的特征点或图像边界上高曲率的点为检测对象,根据灰度或滤波函数确定区域极值点(如H arris角点[4]等),并拓展至不同掩膜下的尺度空间中(如高斯-拉普拉斯、高斯差分等),分析极值点的稳定特性,得到仿射不变的H arris二阶矩描述符[5].线特征描述图像中目标区域的外表形状和轮廓特性,这类轮廓线特征以C anny算子等经典边缘检测算法为基础,集中解决边缘曲线的描述、编组以及组合表达等问题.边缘上的双切线点和高曲率点可连接形成有效的边缘链或圆弧,根据聚类策略或某些规则完成线片段编组,形成线特征的视觉词汇[6-8].区域是图像上具有灰度强相关性的像素集合,包含某种相似属性(如灰度值、纹理等),相对于点线特征,面特征有更丰富的结构信息.区域特征以点特征为中心,采用拉普192模式识别与人工智能23卷拉斯尺度下的H arris或H essian仿射区域描述,对特征尺度上的椭圆仿射区域内的初始点集进行参数迭代估计,根据二阶矩矩阵的特征值测量点邻的仿射形状[4,9].另一种策略分析视觉显著区域对象(如直方图、二值分割图等)的熵值统计特性,得到最佳尺度下的最稳定区域,满足视觉词汇的高重用性[10-11].鲁棒特征表达对提取的特征进行量化表示.点特征一般仅具有图像坐标.线特征则充分考虑邻域边缘点的上下文形状特性,以边缘上采样点为圆心,在极坐标下计算落入等距等角间隔区域的边缘像素直方图.椭圆形面特征描述主要以尺度不变特征变换(Sca le I nvariant Fea t u re Transfor m,SI FT)[12-13]为主,SI FT特征对每个高斯窗口区域估计方向直方图,选择峰值作为参考方向基准,计算4@4网格区域内8个方向的梯度直方图,任何区域均可转换为4@4@8 =128维特征向量.该特征对图像尺度、旋转具有不变性,对亮度和视角改变也保持一定稳定性.通过对特征向量的聚类,得到最原始的特征词汇,形成的语义化图像特征也称为/码书0(Codebook)[14].2.2图像语义的上下文表达图像的语义信息描述主要包含外观位置信息和上下文信息,前者如2.1节所述,可表示成/码书0.上下文信息不是从感兴趣的目标外观中直接产生,而来源于图像邻域及其标签注解,与其他目标的外观位置信息密切相关.当场景中目标外观的可视程度较低时,上下文信息就显得尤为重要.B ieder m an将场景中不相关目标关系分为5种,即支撑(Support)、插入(I nterpositi o n)、概率(Proba-b ility)、位置(Positi o n)和大小(Size)[15-16].五类关系均包含/知识0,不需要知道目标信息就可确定支撑和插入关系,而后三类关系对应于场景中目标之间的语义交互关系,可缩短语义分析时间并消除目标歧义,通常称为/上下文特征0(C ontex t Features),譬如一些相对复杂的特征描述(如全局G ist特征[17-18]、语义掩码特征等)融入场景上下文信息,本身就包含语义(关联)信息,是语义分析的基础.如今有很多研究开始挖掘B ieder m an提出的三类语义关系,可分为语义上下文、空间上下文和尺度上下文[19].语义上下文表示目标出现在一些场景中,而没有出现在其他场景中的似然性,表示为与其他目标的共生(Co-O ccurrence)关系,可采用语义编码方式[20-21],也可由共生矩阵判断两类目标是否相关[22-23],此类上下文对应B ieder m an关系中的/概率0关系.空间上下文表示目标相对于场景中其他目标出现在某个位置上的似然性,对应于/位置0关系.空间上下文隐式地对场景中目标的/共生0进行编码,为场景结构提供更加具体的信息,只需确定很少的目标,就可通过合理的目标空间关系降低目标识别的误差,消除图像中的语义歧义[24-25].尺度上下文表示目标在场景中可能的相对尺度范围,对应于/大小0关系.尺度上下文需处理目标之间的特定空间和深度关系,可缩小多尺度搜索空间,仅关注目标可能出现的尺度.尺度上下文在二维图像中较为复杂,目前仅用于简单的视觉分析系统中[26-27].目前大多数上下文方法主要分析图像中的语义上下文和空间上下文.语义上下文可从其他两种上下文中推理获取,与场景中的目标共生相比.尺度和空间上下文的变化范围较大,而共生关系的知识更易获取,处理计算速度更快.融入上下文特征的图像语义形成了全局和局部两种分析策略,即基于场景的上下文分析和基于目标的上下文分析.前者从场景出发[15,27],将图像统计量看作整体,分析目标和场景之间的高频统计特性,获取全局上下文信息,如马路预示着汽车的出现.后者从目标出发[25,28],分析目标间的高频统计特性,获取局部上下文信息,如电脑预示着键盘的出现.总之,上下文特征包含了更丰富的知识,有助于为图像理解提供更准确的语义信息.2.3语义分析的生成方法生成方法基于模型驱动,以概率统计模型和随机场理论为核心,遵循经典的贝叶斯理论,定义模型集合M,观察数据集合D,通过贝叶斯公式,其模型后验概率p(M|D)可以转换为先验概率p(M)和似然概率p(D|M)的乘积.生成方法一般假设模型遵循固定的概率先验分布(如高斯分布等),其核心从已训练的模型中/生成0观察数据,测试过程通过最大似然概率(M ax i m ize L i k e lihood)得到最符合观察数据分布的模型预测似然(Pred icti v e Like li h ood).图像语义分析的生成方法直接借用文本语义分析的图模型结构(G raph ica lM ode ls),每个节点定义某种概念,节点之间的边表示概念间的条件依赖关系,在隐空间(Latent Space)或随机场(Rando m Field)中建立文本词组和视觉描述之间的关联,生成方法无监督性明显,具有较强的语义延展性.2.3.1层状贝叶斯模型图模型的节点之间由有(无)向边连接,建立视觉词汇和语义词语之间的对应关系.朴素贝叶斯理论形成的经典Bags-o-f W ords模型是层状贝叶斯模1932期高隽等:图像语义分析与理解综述型的雏形,该模型将同属某类语义的视觉词汇视为/包0,其图结构模型和对应的视觉关系描述如图1(a)所示,其中灰色节点为观察变量,白色节点为隐变量,N 为视觉词汇的个数,通过训练建立类别语义描述c 和特征词汇w 之间的概率关系,选取最大后验概率p (c |w )对应的类别作为最终识别结果.(a)朴素贝叶斯(b)概率隐语义分析(c)隐狄利克雷分配(a)N a Çve bay es(b)P robab ili stic latent se m antic ana l y si s (c)L atent D irich let a llocati on图1 有向图语义描述F i g .1 Se m antic i nterpre tati on of directed g raphs朴素贝叶斯模型试图直接建立图像和语义之间的联系,但由于视觉目标和场景的多样性导致这种稀疏的离散分布很难捕捉有效的概率分布规律,因此H o f m ann 借鉴文本分析中的概率隐语义分析(Probab ilistic Latent Se m antic Ana l y sis ,pLSA )模型[29-30],将/语义0描述放入隐空间Z 中,生成相应的/话题0(Top ic)节点,其基本描述如图1(b )所示.D 为M 个图像d 组成的集合,z 表示目标的概念类别(称为/Top ics 0),每幅图像由K 个Topics 向量凸组合而成,通过最大似然估计进行参数迭代,似然函数为p (w |d )的指数形式,与语义词汇和图像的频率相关.模型由期望最大化(E xpec ta tion M ax i m ization,E M )算法交替执行E 过程(计算隐变量后验概率期望)和M 过程(参数迭代最大化似然).决策过程的隐变量语义归属满足z*=arg m ax z P (z |d ),pLSA 模型通过隐变量建立特征与图像间的对应关系,每个文本单元由若干个语义概念按比例组合,本质上隐空间内的语义分布仍然是稀疏的离散分布,很难满足统计的充分条件.隐狄利克雷分配(LatentD ir ich let A llocation ,LDA )模型[31-32]在此基础上引入参数H ,建立隐变量z 的概率分布.在图像语义分析中,变量z 反映词汇集合在隐空间的聚类信息,即隐语义概念,参数H (通常标记为P )则描述隐语义概念在图像空间中的分布,超参A (通常标记为c)一般视为图像集合D 中已知的场景语义描述.如图1(c )所示,由参数估计和变分(V aria tiona l)推理,选取c =arg m ax c P (w |c ,P ,B )作为最终结果.LDA 中不同图像场景以不同的比例P 重用并组合隐话题空间全局聚类(G l o ba lC l u ster),形成/场景-目标-部分0的语义表达关系.LDA 中的隐话题聚类满足De Finetti 可交换原理,其后验分布不受参数次序影响,不同隐话题聚类相互独立,无明显的结构特性.一种显而易见的策略就是在此模型基础上融入几何或空间关系,即同时采用话题对应的语义化特征的外观描述和位置信息,这样不同话题的分布大体被限定于图像场景的某个区域,如天空总是出现在场景的上方等,减小模型决策干扰.如L i 等人[14,33]在LDA 模型中融入词汇的外观和位置信息,并将语义词汇描述c 划分为视觉描述词汇(如sky )和非视觉描述词汇(如w i n d)两类,由词汇类别转换标签自动筛选合适的词汇描述.模型采用取样(Sa mp li n g)策略对从超参先验中生成的视觉词汇和语义标签进行后验概率学习,模型中包含位置信息的语义特征显式地体现了空间约束关系,具有更好的分析效果.(a)无结构(b)全互连结构(c)星状结构(a)U nstructured(b)Fu ll structure (c)Sta r struct u re图2 Part -based 模型表示图F i g.2 R epresen tati on for Part -based m ode lsLDA 模型已明确地将隐空间的/话题0语义进行合理聚类,建立与视觉词汇聚类的对应关系.隐话题聚类隐式地对应场景或目标的某些部分(parts),是一种较原始的par-t based 模型.真正的par-t based模型侧重/目标-部分0之间的语义关联表达,不仅具有较强的结构特性,而且直接概念化隐空间的语义聚类,每个part 直接显式对应语义描述(如人脸可分为眼睛、鼻子、嘴等不同部分).如图2所示,一般通过人工设定或交叉验证的方式固定重要参数(如隐聚类个数、part 个数等)并混合其概率密度,其中固定参数的D ir i c h let 生成过程是一种有限混合./星群0(Conste llati o n)模型[34-35]是其中的典型,根据不194模式识别与人工智能 23卷同区域的外观位置信息描述,确定P 个部分的归属及其概率分布,将目标和背景似然比分解为外观项、形状项、尺度项以及杂项的乘积,依次计算概率密度值(一般是高斯分布或均匀分布),并E M 迭代更新参数,最后通过似然比值判断目标的语义属性.部分间的约束关系体现于形状项中,可以假设为全互连结构(Fu ll Str ucture)或星状结构(S tar S tructure),其结构信息体现于高斯分布的协方差矩阵中(满秩或稀疏矩阵),有助于提高语义分析的准确性.固定参数的D irichlet 生成过程是无限混合模型的一种特例,可通过合适的随机过程,很好表达无限混合(I nfi n ite M i x t u re)模型,自动确定混合个数.这种/非参0(Non -Para m etric)模型可捕捉到概率空间的隐性分布,不受特定的概率密度函数形式表达限制.整个D irich let 过程可拓展至层次结构(H ierar -ch ical D irichlet Process ,HDP).H DP 具有明显的结构特性,可以很容易对应于图像中的/场景-目标-部分0层次结构,其混合组成很显式地表达了不同目标实体间的语义包含关系.Sudderth 在HDP 的基础上,引入转换函数(Transfor m ed Function),生成转换D irichlet 过程(T ransfor m ed D irichlet Process ,TDP),每组的局部聚类不再直接/复制0全局聚类参数,而是通过不同转换函数生成变化多样的局部变参,更符合目标多变特性[36-37].层状贝叶斯模型是当前处理图像语义问题的关注热点,其模型特有的参数化层次结构信息参照文本处理直接对应图像中的语义实体,通过图模型的参数估计和概率推理得到合适的语义描述.模型本身的发展也具有一定的递进关系,即/Bags -o-f W ord模型y pLSA 模型y LDA 模型y par-t based 模型y HDP 模型y TDP 模型0等,分析得到的结果具有层次语义包含关系.2.3.2 随机场模型随机场模型以均值场(M ean F ield)理论为基础,图中节点变量集合{x i |i I V }通常呈4-邻域网格状分布,节点之间的边{(x i ,x j )|i ,j I V;(x i ,x j )I E }体现隐性关联,由势函数W ij (x i ,x j )表示,一般具有含参数H 的近高斯指数分布形式,每个隐节点x i 一般对应一个观察变量节点y i ,由势函数W i (x i ,y i )表示.如图3所示,观察节点可对应图像的像素点,也可对应图像中的某个区域或目标语义化特征描述(如2.1节所述),隐变量则对应语义/标记0或/标签0l .随机场模型具有丰富的结构场信息,节点间上下文关联很强,通常分析像素标记解决图像分割问题.近年来,其特定的约束关系(如桌子和椅子经常关联出现)也被用于图像区域化语义分析中,隐节点集的语义标签对应不同的语义化特征和势函数取值,最大化随机场的能量函数得到的标记赋值,就是最终的区域语义标记属性.随机场模型具有较成熟的计算框架,融合其上下文关联信息的层次贝叶斯/生成0模型是分析图像语义的主流趋势[14,33-35,38-40].图3 随机场模型及其图像语义描述F ig .3 R andom field m ode l and its se m antic descr i pti on2.4 语义分析的判别方法判别方法基于数据驱动,根据已知观察样本直接学习后验概率p (M |D ),主要通过对训练样本的(弱)监督学习,在样本空间产生合适的区分函数,采用形成的分类器或结构参数,完成对特定的特征空间中点的划分(或闭包),形成某些具有相似特性的点的集合.这些共性可直接显式对应图像理解中的若干语义信息,如目标和场景的属性、类别信息等,通常以主观形式体现于观察样本中,其本质就在于学习并获取区分不同语义信息的知识规则(如分类器等).由于语义信息主观设定(如判别几种指定类别),因此判别方法主要侧重观察样本(语义)的处理分析,而非观察样本(语义)的获取.判别方法是包含经典的机器学习方法,精确度较高且易于实现,常用于目标检测识别识别.其策略主要包括最近邻分析、集成学习和核方法.2.4.1 最近邻方法最近邻(k -N earestN e ighbo r ,kNN )方法是基于样本间距离的一种分类方法.其基本思想是在任意空间中、某种距离测度下,寻找和观测点距离最接近的集合,赋予和集合元素相似的属性集合.在图像理解中,就是在图像特征空间寻找和近似的特征描述集,将已知的语义作为分析图像的最终结果.最近邻方法非常简单,但对样本要求较高,需要很多先验知1952期 高 隽 等:图像语义分析与理解综述识,随着大规模语义标记图像库的出现(如后 3.2节所述),最近邻方法有了广阔的应用前景,Torra l b a 等人[41]建立80万幅低分辨率彩色图像集合和相应的语义标记,图像集涵盖所有的视觉目标类别,以W ord N et语义结构树(如后3.1节所述)的最短距离为度量,采用最近邻方法分别对其枝干进行投票,选取最多票数对应最终的语义标签输出.也可直接在图像空间中计算像素点的欧式距离,得到与分析图像相类似的语义空间布局(Con fi g uration).Russe ll 等人[42]利用最近邻方法找出与输入图像相似的检索集,通过含有标记信息的检索图像知识转化到输入图像中,完成场景到目标的对齐任务.语义聚类法还被用于视频数据库中[43],具有较好的结果.2.4.2集成学习集成学习将各种方法获得的模型在累加模型下形成一个对自然模型的近似[44-45],将单一学习器解决问题的思想转换为用多个学习器来共同解决问题.Boosti n g是集成学习方法的典型.其基本思想是每次迭代t生成一个带权重A t的弱分类器(W eaker C lassifier)h t,加大误分样本的权重,保证后续学习对此类样本的持续关注,权重A t表示该弱分类器h t 的重要性,分类效果好的权重大,效果差的权重小.其集成学习的结果就是弱分类器的加权组合E T t=1Ex i I DA t h t(x i)构成一个分类能力很强的强分类器(Strong C lassif-i er),完成简单的二值或复杂的多值分类[46-47].集成学习方法经常用于图像理解的语义分类中,其样本数据集既可以是区域块也可以是滤波后的基元乃至包括上下文和空间布局信息.其分类结果具有很明显的语义区分度.多语义分类中经常出现多类共享的情况,因此,联合Boosti n g的提出极大地减少了分类器的最佳参数搜索时间,使单一弱学习器具有多类判别能力[48-51].同时,近年来多标签多实例(M ult-i Instance M u lt-i Labe l Learn i n g,M I M L)的集成学习策略[52]也倍受学者关注,图像理解中的语义划分问题可通过M I M L转化为单纯数据下的机器学习问题,其输出的分类结果就是对既定语义的编码结果.2.4.3核方法核方法(Kernel)是在数据集中寻找合适的共性/基0,由/基0的混合组成共性空间,与图像理解中的低层基元表示异曲同工.使用核方法可将低维输入空间R n样本特征映射到高维空间中H,即5B R n y H,将非线性问题转换为线性问题.其关键是找到合适的核函数K保持样本在不同空间下的区分关系,即K(x i,x j)=5(x i)#5(x j).它能够在学习框架和特定知识之间建立一种自然的分离来完成图像有意义的表达[53-54].支持向量机(S VM)是常用的核方法之一.它以训练误差作为优化问题的约束条件,以置信范围值最小化作为优化目标,在核函数特征空间中有效训练线性学习分类器,通过确定最优超平面(H yper Plane)及判别函数完成高维空间点的分类.SVM方法在解决小样本、非线性及高维模式识别问题中表现出许多特有的优势,在图像理解中,能有效解决不同环境、姿态以及视角下的广义目标识别分类问题,是目前最为通用的分类模型[55-58].针对多语义分类问题,Farhad i等人[59]将目标的语义属性细分为部分、形状及材质等,相同或相似的语义对应的样本集表明了某种特有的共性关系,采用L1测度对数回归和线性SVM方法学习不同语义类别的判别属性,其多语义属性的不同划分决定了指定目标的唯一描述,具有很强的语义可拓展性.判别模型是通过模型推理学习得出的后验概率,对应不同类别目标的后验概率或对应图像前景和背景的不同后验概率来划定判决边界,进而完成目标识别,指导图像理解.判别模型在特征选取方面灵活度很高,可较快得出判别边界.2.5图像句法描述与分析人对图像场景理解的本质就是对图像本身内在句法(G ra mm ar)的分析.句法源于对语句结构研究,通过一系列的产生式规则将语句划分为相互关联的若干词汇(组)组合,体现句法内词汇之间的约束关系.图像句法分析直接研究图像语义,随着20世纪70年代句法模式识别的提出,Otha就试图构建统一的基于视觉描述的知识库系统,利用人工智能相关策略进行场景语义推理.但由于视觉模型千变万化,方法针对性很强,句法分析方法曾一度没落.当前图像语义分析的一部分研究重心又重新转向图像句法.由于句法分析本身已较为成熟,因此如何建立和句法描述相对应的图像视觉描述非常关键.2.5.1图像与或图表达图像I内的实体具有一定的层次结构,可用与或图(And-O r G raph)的树状结构表示,即解析树pg.如图4所示,同属一个语义概念的实体尽管在外观上具有很大差异,但与或图表达相似,与节点表示实体的分解(D ecom position),如/场景y目标0, /目标y部分0等,遵循A y BCD,的句法规则,或节点表示可供选择的结构组成,遵循A y B|C|D,196模式识别与人工智能23卷。
2001考研英语阅读真题及详细解析Part OneSpecialisation can be seen as a response to the problem of an increasing accumulation of scientific knowledge. By splitting up the subject matter into smaller units, one man could continue to handle the information and use it as the basis for further research. But specialisation was only one of a series of related developments inscience affecting the process of communication. Another was the growing professionalisation of scientific activity.No clear-cut distinction can be drawn between professionals and amateurs in science: exceptions can be found to any rule. Nevertheless, the word 'amateur' does carry a connotation that the person concerned is not fully integrated into the scientific community and, in particular, may not fully share its values. The growth of specialisation in the nineteenth century, with its consequent requirement of a longer, more complex training, implied greater problems for amateur participation in science. The trend was naturally most obvious in those areas of science based especially on a mathematical or laboratory training, and can be illustrated in terms of the development of geology in the United Kingdom.A comparison of British geological publications over the lastcentury and a half reveals not simply an increasing emphasis on the primacy of research, but also a changing definition of what constitutesan acceptable research paper. Thus, in the nineteenth century, localgeological studies represented worthwhile research in their own right; but, in the twentieth century, local studies have increasingly become acceptable to professionals only if they incorporate, and reflect on, the wider geological picture. Amateurs, on the other hand, have continued to pursue local studies in the old way. The overall result has been to make entrance to professional geological journals harder for amateurs, a result that has been reinforced by the widespread introduction of refereeing, first by national journals in the nineteenth century and then by several local geological journals in the twentieth century. As a logical consequence of this development, separate journals have now appeared aimed mainly towards either professiona l or amateur readership. A rather similar process of differentiation has led to professional geologists coming together nationally within one or two specific societies, whereas the amateurs have tended either to remain in local societies or to come together nationally in a different way.Although the process of professionalisation and specialisation was already well under way in British geology during the nineteenth century, its full consequences were thus delayed until the twentieth century. In science generally, however, the nineteenth century must be reckoned as the crucial period for this change in the structure of science.1. The growth of specialisation in the 19th century might be more clearly seen in sciences such as ________.[A] sociology and chemistry [B] physics and psychology[C] sociology and psychology [D] physics and chemistry2. We can infer from the passage that ________.[A] there is little distinction between specialisation and professionalisation[B] amateurs can compete with professionals in some areas of science[C] professionals tend to welcome amateurs into the scientific community[D] amateurs have national academic societies but no local ones3. The author writes of the development of geology to demonstrate________.[A] the process of specialisation and professionalisation[B] the hardship of amateurs in scientific study[C] the change of policies in scientific publications[D] the discrimination of professionals against amateurs4. The direct reason for specialisation is ________.[A] the development in communication [B] the growth of professionalisation[C] the expansion of scientific knowledge [D] the splitting up of academic societiesUnit 8 (2001) Part 1重点词汇:1.specialisation(专业化)即special+is(e)+ation，special(特别的;额外的)，-ise动词后缀(specialise即v.专业化)，-ation名词后缀;specialist(专家;专科医生)?special+ist后缀表“人”。
Unit 11. What is definition?A definition is a statement which captures the meaning, the use, the properties, the function and the essence of a term, a thing or a concept.2. Why is it important to define things or concepts in a clear way?A clear definition helps to avoid misunderstandings or confusions.3. Why should the abstract concepts be defined?A clear definition is required for abstract concepts, such as ‘happiness’, because the definition may change from person to person.4. How many parts does a formal definition consist of? What are they?A formal definition contains three parts: the term, class and differentiating features.5. What types of definition have you learned? In what way do they mainly differ?One-sentence definition (a formal/naming definition) and an extended definition.They mainly differ in length.6. What are the narrative materials which can be used for an extended definition?An extended definition uses narrative materials such as facts, examples, anecdotes, etc. to interpret a term.7. Why should an extended definition be used?Extended definitions will help you get a concept or theory cross to the audience. Making an extended definition requires you to think more carefully about your intended audience and what they need to know.8. How will you write an extended definition?Step 1. Specify the term being defined and its category or class.Step 2. Present clear and basic information of the differentiating features.Step 3. Expand by adding more specific information.Step 4. Use such narrative materials as facts, examples, or anecdotes that readers can understand better.9. In academic contexts, narrative materials cover a broad scope. What can be used as narrative materials?Descriptions, book reports, proposals, presentation of data, explanation of ideas, illustration of evidence, examples, etc.Unit 21. What is contextualization?The process of introducing background information is called contextualization. In any kind ofwriting, providing background information is essential to familiarizing the reader with thetopic and helping the audience get a fuller picture of the topic.2. Could you list several types of context?The cultural, economic, political, historical, social, philosophical, or technological context.3. Where can you find the contextualization of a research problem in an academicarticle?In the introductory section.4. What kind of plane figure does the structure of an introduction resemble?An inverted triangle.5. Where does contextualization fit within the inverted triangle introduction?DefiningBriefly introducing and explaining the research topicContextualizingPutting the research problem into context and pointingout the gap in knowledge6. What are the different types of introduction?1) Inverted triangle introduction2) Contrast introduction3) Anecdote introduction4) Question/quotation introduction7. What kind of writing is mainly used in introducing background information?Description.Unit 31. What is classification?Classification is the process of arranging items into categories based on shared characteristics. Classification is a convenient way to organize information to make thingsclearer and to avoid confusion.2. A common standard should be employed to rate and rank the categories, and toensure that the categories follow a single organizing principle.3. What is the difference between comparison and contrast?Comparison emphasizes the similarities between things, ideas, concepts, or points of view, while contrast emphasizes the differences.4. What is the difference between comparison and mere description?1)The similarities or differences between things are analyzed in comparison in order tomake connections and generate an interesting analysis.2)Description is to present information, accepted knowledge, observations, process ormethod objectively.Unit 41. What are the possible relationships between two notions?1) a causal relationship2)an inclusive relationship3)an opposite relationship4) a parallel relationship2. How do the different relationships function in explaining a notion?Generally, these relationships will provide a clear outline for writers to present a notion and a clear structure for readers to obtain information from the explanation.3. What is an analogy?An analogy is a comparison between two things, usually for the purpose of explanation or clarification. It aims to explain one thing by comparing it to something that is familiar.4. What are the features of an effective analogy?1)On the basis of an exact or similar idea2)Simplicity3)Clear embodiment of the concept or relationship between things5. What effective tools can be used to draw an analogy?Metaphors and similes are tools used to draw an analogy. However, analogy is more extensive and elaborate than either a simile or a metaphor.6. What should be taken into consideration when explanations are given?Just like the targeted audience should be considered while you define new concepts or ideas, you should also consider contexts and situations.7. What are the suggested steps to achieve a logical structure of explanation?1)Clarify your thinking by writing down all the necessary and important details.2)Structure those details logically by deciding which should be explained first andwhich later.Unit 51. What is a summary?A summary is a condensed version of the original text or lecture.2. What is the purpose of a summary?The purpose of a summary is to give the reader a clear, objective picture of the original text.3. The summary of a research paper is found in its Abstract（摘要）.4. In general an abstract includes: objective/aim/purpose; design/methodology; results/findings; conclusions5. When reading a book, one can situate and evaluate the book by reading its publishing details（出版信息）, and navigate the book by reading its table of contents（目录）6. The three reading strategies helping readers to grasp the main idea of an academic text:Skimming and scanning （跳读和略读/浏览）Intensive reading （精读/细读）Analysis of cohesion （分析衔接/信息之间的逻辑关系）7. When you write summaries, you should follow some conventions. They are:1)Quote selectively (selectively or extensively? )2)When you quote, use quotation marks and document the quotation. Failure to do sois plagiarism3)Use present (present or past?) tense to summarize.4)Use summarizing language, for example : the article claims (provide one example)8. What criteria can be adopted to evaluate a summary?1)Accurately and objectively represents the author’s central claim and keysupporting details.2)The summary is not merely listing the main ideas, but show how the reasons supportthe central claim.9. What are the four steps to summarize an article?1)Read the original text intensively to grasp its meaning.2)Note down the main idea and the major supporting ideas.3)Draft your summary, using a mix of paraphrased and quoted material from the text.4)Revise the draft to meet the requirements (Length, grammar, format, documentation,etc.)Unit 61. What is synthesis?It is the act of combining separate things into a coherent whole. It involves analysis related to classification and division, comparison and contrast. It makes use of the ideas of other people, combining sources into one's own words in order to further understanding or establish context.2. What is key to a synthesis?Searching for the flaws, weaknesses, or limitations, and any potential links between various sources, is key to a synthesis.3. What is the difference between a summary and a synthesis?Summary: A summary is a recap or restatement of the important information of the source. The ideas, information and arguments of each source are stated in a concise manner. Synthesis: A synthesis critically analyzes and evaluates the information, including a critical analysis of the relationship between different sources, relate the sources to the author’s own research.Unit 71. What is a literature review?A literature review is a critical survey of important articles, books and other primary sources related to a research topic.2. What are the four ways of organizing a literature review?Literature reviews are often organized in one of these ways: chronologically, thematically, methodologically, or in a combination of these ways.3. What should an author do to make a chronological review?In a chronological review, the author groups and discusses selected sources in order of their publication, highlighting the changes in research over time.4.What should an author do to make a thematic review?In a thematic review, you group and discuss your sources in terms of the themes, theoretical concepts, findings, and topics.5. In what sequence should a thematic literature review proceed?The sequence of the concepts or themes should proceed from the broad to the specific.Unit 81. What is a report?A report is a logical and well-structured piece of writing that describes and analyses a particular subject or problem, communicates information collected from research or the analysis of data.2.What are the differences between a report and an essay?1) A report is fact-based and aim to convey information, while an essay is idea-basedand is written to discuss different opinions and arguments;2) A report is broken into sections and subsections, while an essay usually flow as acontinuous text;3) A report often includes graphics which rarely appears in an essay;4) A report may make predictions or recommendations while an essay does n’t.3.How many types of reports can you list?1)Scientific reports;2)Technical reports;3)Business reports;4)Field reports.4.What sections does a report normally have?1) A title page;2)An abstract;3)An introduction;4)Methods;5)Results;6)Discussion;7)Conclusion;8)Recommendations;9)References;10). Appendices.。
BibliographyALEXANDER, S. (1995). Teaching and learning on the World Wide Web. AUSWEB95 -The first Australian World Wide Web Conference, 30 April - 2 May 1995. On-line: .au/aw95/education2/alexander/index.html (accessed:3 October 2001).ANDERSON, T. (2001). The hidden curriculum in distance education. Change 33(6):28-35.ANON. (2001a). Concepts: Telematic education. Department of Telematic Learning and Education Innovation, University of Pretoria. On-line:http://www.up.ac.za/telematic/concepts/telematic.htm (accessed: 12 September 2001).ANON. (2001b). What Is Contextual Learning? On-line:/lev2.cfm/56(accessed: 13 November 2001)ÅKERLIND, G. & TREVITT, C. (1995). Enhancing learning through technology: when students resist the change. ASCILITE 95 - Learning with Technology, 3-7 December, Melbourne, Australia. On-line:.au/conferences/melbourne95/smtu/papers/akerlind.pdf (accessed: 20 August 2001).ARNOLD, M. (1997). Using the Web to augment teaching and learning. ASCILITE 97 -What works and why, 7-10 December, Perth, Australia. On-line:.au/conferences/perth97/papers/Arnold/Arnold.html(accessed: 24 September 2001).BAIRD, J.R. & MITCHELL, I.J. (1986). Improving the quality of teaching and learning: An Australian case study - The Peel Project. Melbourne: Peel Group, Monash University. BARRON, A.E. (1998). Designing Web-based training. On-line:/itforum/paper26/paper26.html (accessed: 6 November 2001).BASCH, R. (1998). Web publishing: A voice for the people. Smart Computing: Guide Series 6(6). On-line: (accessed: 21 September 2001).BASCH, R. (1999). How to discover information online. Smart Computing: Learning Series 5(3). On-line: (accessed: 21 September 2001).BELKIN, N.J. (1975). Towards a definition of information for informatics. In: V. Horsnell (Ed.), Informatics 2: 50-56. London: Aslib.BERENSON, M.L. & LEVINE, D.M. (1986). Basic business statistics: Concepts and applications. Engelwood Cliffs: Prentice-Hall International.BLOOM, B.S., ENGELHART, M.D., FURST, D.J., HILL, W.H. & KRATHWOHL, D.R.(1956). Taxonomy of educational objectives: The classification of educational goals.New York: David McKay.BOHLIN, R.M. (1999). Avoiding computer avoidance. On-line:/itforum/paper35/paper35.html (accessed: 24 September 2000).BRANSFORD, J.D., BROWN, A.L. & COCKING, R.R. (1999). How people learn: Brain, mind, experience and school. Washington: National Academy Press.BROOKES, B.C. (1975). The fundamental problem of information science. In: V. Horsnell (Ed.), Informatics 2: 42-49. London: Aslib.BROOKES, B.C. (1977). The developing cognitive viewpoint in information science. In: CC-77: Int Workshop on the Cognitive Viewpoint. Ghent: Ghent University. BROOKES, B.C. (1980). The foundations of information science: Part 1: Philosophical aspects. Journal of Information Science, 2: 125-133.BROOKES, B.C. (1981). The foundations of information science: Part 4: Information science: the changing paradigm. Journal of Information Science, 3: 3-12.BROWN, T. (2000). (tbrown@postino.up.ac.za) Pers. Comm.BUCKNER, T. (terribuckner@) (2001). Re: my 2 cents plus another 2.Email to: ITForum listserver (ITForum@). 16 February. BUNDERSON, C.V. & INOUYE, D.K. (1987). The evolution of computer-aided educational delivery systems. In R.M. Gagné (Ed.) Instructional technology:Foundations. New Jersey: Lawrence Erlbaum Associates Inc.CANDY, P.C. (1991). Self-direction for lifelong learning. San Francisco: Jossey-Bass Inc. CARR, B.A. (alan@postino.up.ac.za) (2001a). Re: question. Email to: ITForum listserver (ITForum@). 16 February.CARR, B.A. (2001b). Who is Grace Murray Hopper? (as yet unpublished).CERN. (1997). History and growth [of the World Wide Web]. On-line:http://public.web.cern.ch/Public/ACHIEVEMENTS/WEB/history.html (accessed: 17 September 2001).CLARK, R. (richard_clark2@) (2001). Re: question. Email to: ITForum listserver (ITForum@). 16 February.CLARK, R.E. (1983). Reconsidering research on learning from media. Review of Educational Research, 53: 445-459.CLARK, R.E. (1994). Media will never influence learning. Educational Technology Research and Development, 42(2): 21-29.CLARKE, P.A. (1998). Telematic Teaching of adults via the World Wide Web: A university case study. MEd thesis, University of Pretoria, Pretoria. On-line:http://www.und.ac.za/users/clarke/thesis/index.html (accessed: 18 May 2000). COLLINS PLAIN ENGLISH DICTIONARY. (1996). London: HarperCollins.COPLEY, J. (1992). The integration of teacher education and technology: a constructivist model. In Carey, D., Carey, R., Willis, D. & Willis, J. (eds.) Technology and Teacher Education. Charlottesville: AACE: 681.CREBBIN, W. (1995). Learning, knowledge, language and meaning. A paper presented at The Fourth International Literacy and Education Research Network Conference on Learning, Townsville. On-line:.au/~wcrebbin/homepage/learningpaper.html (accessed: 29 September 2000).CREBBIN, W. (1999). How does learning happen? A paper presented at 2nd National Language and Academic Skills Conference, Monash University, Melbourne. On-line: .au/~wcrebbin/homepage/learningpaper99.html (accessed:29 September 2000).CRONJÉ, J.C. (1997). Interactive Internet: Using the Internet to facilitate co-operative distance learning. South African Journal of Higher Education, 11(2): 149-156. On-line: http://hagar.up.ac.za/catts/abc/rbo96.html (accessed: 24 September 2000). CRONJÉ, J.C. (2001). (jcronje@postino.up.ac.za). Pers. Comm.CRONJÉ, J.C. & CLARKE, P.A. (1998). Teaching 'Teaching on the Internet' on the Internet. On-line: http://hagar.up.ac.za/catts/abc/clarke&cronjec.doc (accessed:24 September 2000).DAVENPORT, T.H. (with PRUSAK, L.) (1997). Information Ecology: Mastering the Information and Knowledge Environment. New York: Oxford University Press. DAVENPORT, T.H. & PRUSAK, L. (1998). Working Knowledge. Boston: Harvard Business School.De BONO, E (2000). New thinking for the new millenium. Penguin. (first published in 1999 by Viking).De LANGE, A.M. (amdelange@gold.up.ac.za) (2000). Information and Knowledge. Email to: Carr, BA (alan@gold.up.ac.za) 21 August. (Included in this document as Appendix 3.) DEPARTMENTS OF EDUCATION AND COMMUNICATIONS (2001). Strategy for information and communication technology in education. Report by the South African government. On-line:.za/teli2/ICTStrategy November 2001.pdf (accessed:30 November 2001).De VILLIERS, G.J. (2001a). (devilliersg@postino.up.ac.za) Pers. Comm.De VILLIERS, G.J. (2001b). Evaluating asynchronous Web-based technologies that support learning. M.A. thesis, University of Pretoria, Pretoria.DERVAN, B. & NILAN, M. (1986). Information needs and uses. Annual Review of Information Science and Technology. 21: 3-33.DRAPER, S.W. (1997). Adding (negotiated) learning management to models of teaching and learning. ITForum Paper 21. On-line:/itforum/paper21/paper21.html (accessed: 25 September2000).DRAPER, S.W. (1998). Niche-based success in CAL. Computers and Education. 30: 5-8.On-line: /~steve/niche.htmlDRAPER, S.W. (s.draper@psy.gla.ac.za) (2000). Henryk's paper 42. Email to ITForum listserver (itforum@) 7 May.DRAPER, S.W. (s.draper@psy.gla.ac.za) (2001a). learning objectives summary. Email to: ITForum listserver (ITForum@). 21 February.DRAPER, S.W. (2001b). Why show learners the objectives? On-line:/~steve/lobjs.html (accessed: 6 August 2001).ELLIOTT, M. (2001). It's what's behind the economy stupid. Time Magazine (European Edition). 30 July: 34.ERASMUS, L. (2000). (erasmus@) Pers. Comm.ESTES, F. & CLARK, R.E. (1999). Authentic educational technology: The lynchpin between theory and practice. Educational Technology 39(6): 5-13.FLANAGAN, A. (TFlanaga@tsamail.trsa.ac.za) (2000). Re: Literature Review. Email to: UPGuide listserver (upguide@). 10 July.FLANAGAN, A. (TFlanaga@tsamail.trsa.ac.za) (2001). Re: question. Email to: ITForum listserver (ITForum@). 16 February.FLEMING, M.L. (1987). Displays and communication. In R.M. Gagné (Ed.) Instructional technology: Foundations. New Jersey: Lawrence Erlbaum Associates Inc. FORSYTH, I. (1998). Teaching and learning materials on the Internet. London: Kogan Page.FOX, D. (1983). Personal theories of teaching. Studies in Higher Education, 8(2):151-163. GAGNÉ, R.M. (1965). The conditions of learning. New York: Holt, Rhinehart & Winston. GAGNÉ, R.M. (1987). Introduction. In R.M. Gagné (Ed.) Instructional technology: Foundations. New Jersey: Lawrence Erlbaum Associates Inc.GAGNÉ, R.M. & GLASER, R. (1987). Foundations in learning research. In R.M. Gagné(Ed.) Instructional technology: Foundations. New Jersey: Lawrence ErlbaumAssociates Inc.GALUSHA, J.M. (1997) Barriers to learning in distance education. IPCT-J. 5(3-4): 6-14.On-line: /~ipct-j/1997/n4/galusha.html (accessed:11 February 2000).GORMAN, R.W. (1999). KnCell Technologies. On-line: (accessed:17 June 2001).GORMAN, R.W. (bgorman@) (2001a). Re: A More Fundamental Question... .Email to LogoForum listserver (logoforum@) 16 April.GORMAN, R.W. (bgorman@) (2001b). Re: A More Fundamental Question... .Email to LogoForum listserver (logoforum@) 17 April.GORMAN, R.W. (bgorman@) (2001c). Re: Knowledge. Email to: Carr, B.A.(alancarr@) 17 June.GROMOV, G.R. (2000). The roads and crossroads of Internet history. Part 4: Birth of the Web. On-line: /intvalweb.html (accessed: 17 September 2000).HARMON, S.W. & JONES, M.G. (1999). The five levels of Web use in education: Factors to consider in planning online courses. Educational Technology 39(6):28-32. HARP, S.F. & MAYER, R.E. (1998). How seductive details do their damage: A theory of cognitive interest in science learning. Journal of Educational Psychology, 90(3):414-434.HART, G. & GILDING, A. (1997). Virtual tutorials, virtual lectures, virtual prisons.ASCILITE 97 - What works and why, 7-10 December, Perth, Australia. On-line:.au/conferences/perth97/papers/Hart/Hart.html (accessed:24 September 2001).HERSELMAN, M.E. (1999). The application of computer games in English second language teaching. PhD. thesis, University of Pretoria, Pretoria.HONEYBALL, J. (2000). Epilog: Thin pipes. PC Pro Magazine. On-line: /alancarr/php3/openframe.php3?page=column2.html(accessed: 31 August 2000. Note that this is a portal site and this URL refers to a new article each week.)HORAK, M. (1996). Pers. comm. (Mrs Horak can be reached throughemileh@global.co.za)HOUSEGO, S. & FREEMAN, M. Case studies: Integrating the use of web based learning systems into student learning. Australian Journal of Educational Technology.16(3):258-282.HILL, J.R. (2001). Building community in Web-based learning environments: Strategies and techniques. AusWeb01 - The seventh Australian World Wide Web Conference, 21 -25 April 2001. On-line:.au/aw01/papers/refereed/hill/paper.html (accessed:18 June 2001).IHEP (1999). What's the difference? A review of contemporary research on the effectiveness of distance learning in Higher Education. A report of the Institute for Higher Education Policy. On-line: /Pubs/PDF/Difference.pdf (accessed: 13 September 2001). This document has been included on the attached CD in the folder PDF.IHEP (2000). Quality on the line: Benchmarks for success in Internet-based distance education. A report of the Institute for Higher Education Policy. On-line:/quality.pdf (accessed: 13 September 2001). This document has been included on the attached CD in the folder PDF.INGWERSEN, P. (1992). Information retrieval interaction. London: Taylor Graham. INGWERSEN, P. (1996). Cognitive perspectives of information retrieval interaction: elements of a cognitive IR theory. Journal of Documentation. 52(1):3-50. INGWERSEN, P. (2000a). Lectures in the Information Science honours course (INY 773) at the University of Pretoria. February, 21-24 2000.INGWERSEN, P. (pi@db.dk) (2000b). Re: Comments anyone? Email to: Carr, B.A.(alan@gold.up.ac.za) 29 September.IRANI, T. (2000). If we build it, will they come? The effects of experience and attitude on traditional-aged students' views of distance education. International Journal ofEducational Technology. 2(1). On-line:/ijet/v2n1/irani/index.html (accessed: 17 July 2000). JOHNSTON, C. (2000) Fostering deeper learning. On-line:.au/ecowww/fost.html (accessed: 7 July 2000). JONASSEN, D.H. (1988). Designing structured hypertext, and structuring access to hypertext. Educational Technology. 28(11): 13-16.JONES, B. (bjones1@) (2001). question. Email to: ITForum listserver (ITForum@). 15 February.JONES, V., JO, J.H. & CRANITCH, G. (2000). A study of students' response to WBI within a traditional learning environment. AusWeb2k - Australian World Wide Web Conference, 12-17 June 2000. On-line:.au/aw2k/papers/jones/paper.html (accessed: 3 October 2001).JRI. (jri@) (2001). Re: question. Email to: ITForum listserver (ITForum@). 16 February.KANTROWITZ, B. & McGINN, D. (2000). When teachers are cheaters. On-line: /bcom/original/article/0,5744,8138,00.html (accessed:14 August 2000).KAUFMAN, R. & THIAGARAJAN, S. (1987). Identifying and specifying requirements for instruction. In R.M. Gagné (Ed.) Instructional technology: Foundations. New Jersey: Lawrence Erlbaum Associates Inc.KEARSLEY, G. (1988). Authoring considerations for hypertext. Educational Technology.28(11): 21-24.KEMC Coventry University. (1999) FAQ. On-line:/faq/faq.htm (accessed: 23 August 2000). KOZMA, R.B. (1994). Will media influence learning? Reframing the debate. Educational Technology Research and Development, 42(2): 7-19.LAURILLARD, D. (1993). Rethinking university teaching: A framework for the effective use of educational technology. London: Routledge.LÊ, T. & LÊ, Q.(1999). A Web-based study of students' attitudes towards the Web. ED-MEDIA 99. World Conference on Educational Multimedia, Hypermedia &Telecommunications. Charlottesville, Virginia: AACE. On-line:/le.html (accessed: 20 July 2000).MACFARLANE, A. (1995). Future patterns of teaching and learning. In T. Schiller (Ed.), The changing university. Buckingham: Open University Press.MACHLUP, F. (1962). The production and distribution of knowledge in the United States.Princeton: Princeton University Press.MACHLUP, F. (1980). Knowledge: its creation, distribution, and economic significance.Vol. 1: Knowledge and knowledge production. Princeton: Princeton University Press. McINTYRE, D.R. & WOLFF, F.G. (1998). An experiment with WWW interactive learning in university education. Computers and Education 31:255-264.McSHANE, J. (1991). Cognitive development: An information processing approach.Oxford: Blackwell.MARSH, G.E. (gemarsh@) (2000). Re: paper 45: procrastination. Email to: ITForum listserver (ITFORUM@). 22 September.MARTIN, S.B. (1998). Information technology, employment, and the information sector: Trends in information employment 1970-1995. Journal of the American Society for Information Science, 49:1053-1069.MAYER, R.E., STEINHOFF, K., BOWER, G. & MARS, R. (1995). A generative theory of textbook design: Using illustrations to foster meaningful learning of science text.Educational Technology Research and Development, 43: 31-43.MEINTJIES, F. (2001). Left Brain: Internet's great window of opportunity not yet open to all. Sunday Times: Business Times. 26 August: 32.MEYER, J.C. (2000). (meyerjc@sabs.co.za) Discussions under a tree: An interview conducted in the Kruger National Park. November 2000.MILLS, G. (2000). MSWLogo, an educational programming tool. On-line: /logo.html (accessed: 26 September 2000).MORGAN, R.M. (1987). Planning for instructional systems. In R.M. Gagné (Ed.) Instructional technology: Foundations. New Jersey: Lawrence Erlbaum Associates Inc. MORRISON, G.R. (Morrison.Gary@) (2001). Re: designing objectives into instruction. Email to: ITForum listserver (ITForum@). 19February.NEL, Z. (2001). (znel@xsinet.co.za) Pers. comm.NIELSEN, J. (2000). Designing Web usability: The practice of simplicity. Indianapolis: New Riders Publishing.NORDHOFF, H. (2001). (hnordhoff@postino.up.ac.za) Pers. comm.NORTJE, T. (tnortje@it.up.ac.za) (2001). Re: your bosses don't seem to know. Email to: Carr, B.A. (alancarr@). 14 August.O'MALLEY, P. (2001). Spend more on upgrading of teachers. Sunday Times Letters.12 August: 15.ON-LINE DICTIONARY. (2000). (accessed: 13 August 2000). PARAT, M. (1977). The information economy (vols 1-10). Washington DC: Department of Commerce.PASSERINI, K. & GRANGER, M.J. (2000). A developmental model for distance learning using the Internet. Computers and Education, 34: 1-15.PERKINS, D.N. (1986). Knowledge as Design. New Jersey: Lawrence Erlbaum Associates Inc.POPPER, K.R. (1968). Conjectures and Refutations: The growth of scientific knowledge.New York: Harper Torchbooks.PRETORIUS, C. (2001a). Maths and science: What's the solution? Sunday Times.5 August: 19.PRETORIUS, C. (2001b). Tapping into a new way of learning. Sunday Times.19 August: 19.PRETORIUS, C. (2001c). The great new heroes of the blackboard. Sunday Times.23 September: 15.REIGELUTH, C.M. (1983). Meaningfulness and instruction: Relating what is being learned to what a student knows. Instructional Science, 12: 197-218REW, S. (1997). EM 600: Gagne. On-line:/academic/education/eflt/gagne.html (accessed: 13 April 2000).ROSS, S.M. (1994). Delivery trucks or groceries? More food for though on whether media (will, may, can't) influence learning. Educational Technology Research and Development, 42(2): 5-6.RUSSELL, T.J. (2002). The "No Significant Difference Phenomenon". On-line: http://teleeducation.nb.ca/nosignificantdifference/ (accessed: 28 September 2002). SAARENKUNNAS, M., KUURE, L. & TAALAS, P. (1999). Teacher roles and interaction in Web-based learning environments. On-line:http://wwwedu.oulu.fi/homepage/msaarenk/taykmen3.htm (accessed: 10 July 2000).SARACEVIC, T. (1999). Information science. Journal of the American Society for Information Science, 50: 1051-1063.SCHANK, R.C. (1999). Educational Outrage: Performance recognition. On-line: /edoutrage/edoutrage8.html (accessed: 28 August 2000). SCRECKER, E. (1998). Technology and intellectual property: Who's in control.Academe, 84(3): 13.SHEA, C. (2000). It's come to this. Teacher Magazine. May 2000. On-line: /tm/tmstory.cfm?slug=08profit.h11 (accessed: 28 August 2000).SHEARD, J., POSTEMA, M. & MARKHAM, S. (2000). Paper-based and Web-based resources: What do students value? AusWeb2k - Australian World Wide WebConference, 12-17 June 2000. On-line:.au/aw2k/papers/sheard/paper.html (accessed: 3 October 2001).SHERRY, L. (1996). Issues in distance learning. International Journal of Educational Communications, 1(4): 337-365.SHULMAN, L.S. (1987). Knowledge and Teaching: Foundations of the New Reform.Harvard Educational Review, 57: 1-22.SIMS, (2000). How much information? Executive summary. Report of the School of Information Management Studies, University of California Berkeley. On-line:/research/projects/how-much-info/summary.html (accessed: 19 September 2000).SING, L.S. (1999). Problem-solving in a constructivist environment. Educational Technology and Society. 2(4). On-line:/periodical/vol_4_99/lee_chien_sing.html (accessed: 14 April2000).SONNENREICH, W. (1998). A history of search engines. On-line:/legacy/compbooks/sonnenreich/history.html (accessed: 19 September 2000). [Be wary of this page. It downloads a javascript applet whichremoves the page from the browser's cache and also prevents the page from being printed or saved to the hard disk. It can also cause computer lock-ups.]STEYN, A.B. (dolf@twrinet.twr.ac.za) (2000). Re: [upguide] Re: Tony's english. Email to: UPGuide listserver (upguide@).10 July.STEYN, T.M. (1998). Graphical exploration as an aid to mastering fundamental mathematical concepts: An instructional model for mathematics praticals. M.Ed.thesis, University of Pretoria.TAM, M. (2000). Constructivism, Instructional Design, and Technology: Implications for Transforming Distance Learning. Educational Technology and Society. 3(2). On-line: /periodical/vol_2_2000/tam.html (accessed: 8 May 2000). TANNER, H. (1992). Developing the use of IT within mathematics through action research.Computers and Education. 18: 143-148.TAPSCOTT, D. (2001). The net generation and school. On-line:/edtech/article.taf?_function=detail&Content_uid1=109(accessed: 16 August 2001).TELI (1996). Technology-enhanced learning initiative: Discussion document. Report by the South African government. On-line:.za/teli2/policydocuments/plan1.htm (accessed:4 December 2001).TELI (1997). Technology-enhanced learning initiative: A strategic plan. Report by the South African government. On-line:.za/teli2/policydocuments/discussion1.htm (accessed:4 December 2001).TERRAPIN, (2000). Terrapin: Tools for thinking. On-line: (accessed: 26 September 2000).TIFFIN, J. & RAJASINGHAM, L. (1995). In search of the virtual class: Education in an information society. London: Routledge.TODD, R.J. (1999). Back to our beginnings: information utilization, Bertram Brookes and the fundamental equation of information science. Information Processing and Management, 35: 851-870.VAN HARMELEN, T. (1997). The development of a technology-enhanced education strategy for the University of Pretoria. M.Ed. thesis, University of Pretoria. VERMAAK, M.E. (2000). (madeleine.vermaak@pixie.co.za) Pers. Comm.VERMAAK, P. (2000). JavaLOGO. The preliminary JavaLOGO class has been included on the attached CD in the folder JavaLOGO.VON GLEHN, F.H. (2000). (fvg@xsinet.co.za) Pers. Comm.WARD, M. & NEWLANDS, D. (1998). Use of the Web in undergraduate teaching.Computers and Education. 31: 171-184.WBEC (2000). The power of the Internet for learning: Moving from promise to practice.Report of the Web-based Education Commission. On-line:/upload/wbec/reports/WBECReport.pdf (accessed: 16January 2001). This document has been included on the attached CD in the folder PDF.WIJEKUMAR, K. (2001). What is driving Web-based distance learning environments? On-line: /itforum/paper52/paper52.html (accessed: 25 April 2001).。
