计算机英语下 期末 复习(广外)

前言

广外21级计算机英语下半学期末复习，考试考的内容 462-470  Artificial Intelligence

期末考试题型：

1.名词解释

2.翻译 

（如果有翻译错误/小道消息/未补充的知识点请评论，祝大家期末科科4.0！）

一、Intelligent Agents

1. 名词解释  

1 Artificial Intelligence

Artificial Intelligence is the field of computer science that seeks to build autonomous machines  that can carry out complex tasks without  human intervention.

2.翻译 

1 Artificial Intelligence is the field of computer science that seeks to build autonomous machines—machines that can carry out complex tasks without human intervention. This goal requires that machines be able to perceive and reason.

人工智能是计算机科学的一个领域，旨在打造自主机器——无需人工干预就能执行复杂任务的机器。这一目标要求机器能够感知和推理。

2 Such capabilities fall within the category of commonsense activities that, although natural for the human mind, are proving difficult for machines.

这些能力属于常识活动范畴，虽然对于人类思维来说很自然，但对于机器来说却很难。

3 The result is that work in the field continues to be challenging. In this chapter we explore some of the topics in this vast area of research.

因此，该领域的工作仍然充满挑战。在本章中，我们将探讨这一广阔研究领域的一些主题。

4 The field of artificial intelligence is quite large and merges with other subjects such as psychology, neurology, mathematics, linguistics, and electrical and mechanical engineering.

人工智能领域相当庞大，并与心理学、神经学、数学、语言学、电气和机械工程等其他学科相融合。

5 To focus our thoughts, then, we begin by considering the concept of an agent and the types of intelligent behavior that an agent might exhibit.

为了集中我们的思想，我们首先考虑代理的概念以及代理可能表现出的智能行为类型。

6 Indeed, much of the research in artificial intelligence can be categorized in terms of an agent’s behavior.

事实上，人工智能的大部分研究都可以根据代理的行为进行分类。

7 An agent is a “device” that responds to stimuli from its environment.

代理是一种对环境刺激做出反应的“设备”。

8 It is natural to envision an agent as an individual machine such as a robot, although an agent may take other forms such as an autonomous airplane, a character in an interactive video game, or a process communicating with other processes over the Internet (perhaps as a client, a server, or a peer). 

很自然地，将代理设想为诸如机器人之类的单个机器，尽管代理还可以采用其他形式，例如自主飞机、交互式视频游戏中的角色或通过互联网与其他进程通信的进程（可能是客户端、服务器或对等点）。

9 Most agents have sensors by which they receive data from their environments and actuators by which they can affect their environments.

大多数代理都有传感器，它们通过传感器接收来自环境的数据，还有执行器，它们可以通过执行器影响环境。

10 Examples of sensors include microphones, cameras, range sensors, and air or soil sampling devices. Examples of actuators include wheels, legs, wings, grippers, and speech synthesizers.

传感器的示例包括麦克风、摄像头、测距传感器以及空气或土壤采样设备。执行器的示例包括轮子、腿、翅膀、夹子和语音合成器。

11 Much of the research in artificial intelligence can be characterized in the context of building agents that behave intelligently, meaning that the actions of the agent’s actuators must be rational responses to the data received though its sensors. In turn, we can classify this research by considering different levels of these responses.

人工智能领域的许多研究都可以在构建智能行为代理的背景下进行描述，这意味着代理执行器的动作必须是对通过其传感器接收到的数据的合理响应。反过来，我们可以通过考虑这些响应的不同级别来对此类研究进行分类。

12 The simplest response is a reflex action, which is merely a predetermined response to the input data. Higher levels of response are required to obtain more “intelligent” behavior.

最简单的反应是反射动作，这只是对输入数据的预定反应。需要更高级别的反应才能获得更“智能”的行为

13 For example, we might empower an agent with knowledge of its environment and require that the agent adjust its actions accordingly. The process of throwing a baseball is largely a reflex action but determining how and where to throw the ball requires knowledge of the current environment.

例如，我们可以赋予代理有关其环境的知识，并要求代理相应地调整其动作。投掷棒球的过程在很大程度上是一种反射动作，但确定如何以及在哪里投掷球需要了解当前环境。

14 (There is one out with runners on first and third.) How such real-world knowledge can be stored, updated, accessed, and ultimately applied in the decision-making process continues to be a challenging problem in artificial intelligence.

（一垒和三垒有跑垒员，一名跑垒员出局。）如何存储、更新、访问这些现实世界的知识并最终将其应用于决策过程，仍然是人工智能领域的一个挑战性问题。

15 Another level of response is required if we want the agent to seek a goal such as winning a game of chess or maneuvering through a crowded passageway.

如果我们希望代理实现某个目标（比如赢得一场棋局或者穿过拥挤的通道），则需要另一个级别的响应。

16 Such goal-directed behavior requires that the agent’s response, or sequence of responses, be the result of deliberately forming a plan of action or selecting the best action among the current options.

这种目标导向行为要求代理的响应或响应序列是经过深思熟虑制定的行动计划或从当前选项中选择最佳行动的结果

17 In some cases an agent’s responses improve over time as the agent learns. This could take the form of developing procedural knowledge (learning “how”) or storing declarative knowledge (learning “what”).

在某些情况下，随着代理不断学习，其响应会随着时间的推移而改善。这可以采取开发程序性知识（学习“如何”）或存储陈述性知识（学习“什么”）的形式。

18 Learning procedural knowledge usually involves a trial and error process by which an agent learns appropriate actions by being punished for poor actions and rewarded for good ones. 

学习程序性知识通常涉及反复试验的过程，在此过程中，代理通过对不良行为进行惩罚和对良好行为进行奖励来学习适当的行为。

19 Following this approach, agents have been developed that, over time, improve their abilities in competitive games such as checkers and chess. Learning declarative knowledge usually takes the form of expanding or altering the “facts” in an agent’s store of knowledge.

按照这种方法，代理已经发展起来，随着时间的推移，它们在跳棋和国际象棋等竞争性游戏中的能力不断提高。学习陈述性知识通常采取扩展或改变代理知识库中的“事实”的形式。

20 For example, a baseball player must repeatedly adjust his or her database of knowledge (there is still just one out, but now runners are on first and second) from which rational responses to future events are determined.

例如，棒球运动员必须反复调整自己的知识数据库（仍然只有一个出局，但现在跑垒员在一垒和二垒），以此确定对未来事件的合理反应。 

21 To produce rational responses to stimuli, an agent must “understand” the stimuli received by its sensors. That is, an agent must be able to extract information from the data produced by its sensors, or in other words, an agent must be able to perceive. 

为了对刺激做出合理的反应，代理必须“理解”其传感器接收到的刺激。也就是说，代理必须能够从其传感器产生的数据中提取信息，换句话说，代理必须能够感知。

22 In some cases this is a straightforward process. Signals obtained from a gyroscope are easily encoded in forms compatible with calculations for determining responses. But in other cases extracting information from input data is difficult. 

在某些情况下，这是一个简单的过程。从陀螺仪获得的信号很容易以与确定响应的计算兼容的形式进行编码。但在其他情况下，从输入数据中提取信息很困难。

23 Examples include understanding speech and images. Likewise, agents must be able to formulate their responses in terms compatible with their actuators. 

例如理解语音和图像。同样，代理必须能够根据与执行器兼容的术语来制定响应。

24 This might be a straightforward process or it might require an agent to formulate responses as complete spoken sentences—meaning that the agent must generate speech. 

这可能是一个简单的过程，也可能需要代理将响应制定为完整的口头句子——这意味着代理必须生成语音。

25 In turn, such topics as image processing and analysis, natural language understanding, and speech generation are important areas of research.

反过来，图像处理和分析、自然语言理解和语音生成等主题是重要的研究领域。

26 The agent attributes that we have identified here represent past as well as current areas of research. Of course, they are not totally independent of each other. 

我们在这里确定的代理属性代表了过去和当前的研究领域。当然，它们并不是完全独立的。

27 We would like to develop agents that possess all of them, producing agents that understand the data received from their environments and develop new response patterns through a learning process whose goal is to maximize the agent’s abilities. 

我们希望开发出具备所有这些功能的代理，生产出能够理解从其环境中接收到的数据并通过学习过程开发新的响应模式的代理，其目标是最大限度地发挥代理的能力。

28 However, by isolating various types of rational behavior and pursuing them independently, researchers gain a toehold that can later be combined with progress in other areas to produce more intelligent agents.

然而，通过分离各种类型的理性行为并独立追求它们，研究人员获得了立足点，以后可以将其与其他领域的进展相结合，以产生更智能的代理。

29 We close this subsection by introducing an agent that will provide a context for our discussion in Sections 11.2 and 11.3.

我们通过介绍一个代理来结束本小节，该代理将为我们在 11.2 节和 11.3 节中的讨论提供背景。

30 The agent is designed to solve the eight-puzzle, which consists of eight square tiles labeled 1 through 8 mounted in a frame capable of holding a total of nine such tiles in three rows and three columns ( Figure 11.1).  

该代理旨在解决八字谜题，该谜题由八个标有 1 到 8 的方形瓷砖组成，这些瓷砖安装在一个框架中，该框架能够容纳三行三列中总共九个这样的瓷砖（图 11.1）。

31 Among the tiles in the frame is a vacancy into which any of the adjacent tiles can be pushed, allowing the tiles in the frame to be scrambled. The problem posed is to move the tiles in a scrambled puzzle back to their initial positions (Figure 11.1). 

框架中的瓷砖之间有一个空位，可以将任何相邻的瓷砖推入其中，从而允许框架中的瓷砖被打乱。提出的问题是将打乱的拼图中的瓷砖移回其初始位置（图 11.1）。

32 Our agent takes the form of a box equipped with a gripper, a video camera, and a finger with a rubber end so that it does not slip when pushing something.

我们的代理采用盒子的形式，配备抓手、摄像机和带有橡胶末端的手指，以便在推动某物时不会滑落。

33 When the agent is first turned on, its gripper begins to open and close as if asking for the puzzle. When we place a scrambled eight-puzzle in the gripper, the gripper closes on the puzzle.

当代理首次启动时，它的夹持器开始张开和闭合，就像在请求拼图一样。当我们将一个乱码的八块拼图放入夹持器时，夹持器会闭合到拼图上。

34 After a short time the machine’s finger lowers and begins pushing the tiles around in the frame until they are back in their original positions. At this point the machine releases the puzzle and turns itself off.

过了一会儿，机器的手指放下，开始推动框架中的瓷砖，直到它们回到原来的位置。此时，机器释放拼图并自动关闭。

35 This puzzle-solving machine exhibits two of the agent attributes that we have identified.

这个解谜机器展现出了我们已经识别的两个代理属性。

36 First, it must be able to perceive in the sense that it must extract the current puzzle state from the image it receives from its camera. We will address issues of understanding images in Section 11.2. Second, it must develop and implement a plan for obtaining a goal. We will address these issues in Section 11.3.

首先，它必须能够感知，也就是说，它必须从相机接收到的图像中提取当前的拼图状态。我们将在第 11.2 节中讨论理解图像的问题。其次，它必须制定并实施实现目标的计划。我们将在第 11.3 节中讨论这些问题

二、Research Methodologies 

1. 名词解释  

1 Natural Language Processing (NLP)

A field of AI focused on the interaction between computers and humans through natural language. The goal is for computers to process and analyze large amounts of natural language data.

2.翻译 

1  To appreciate the field of artificial intelligence, it is helpful to understand that it is being pursued along two paths. One is the engineering track in which researchers are trying to develop systems that exhibit intelligent behavior.

要理解人工智能领域，了解其发展路径有两条，会很有帮助。一条是工程轨道，研究人员试图开发表现出智能行为的系统。

2  The other is a theoretical track in which researchers are trying to develop a computational understanding of animal—especially human—intelligence. This dichotomy is clarified
by considering the manner in which the two tracks are pursued. 

另一条是理论轨道，研究人员试图发展对动物（尤其是人类）智能的计算理解。通过考虑这两种轨道的实现方式，可以澄清这种二分法。

3  The engineering approach leads to a performance-oriented methodology because the underlying goal is to produce a product that meets certain performance goals.  

工程方法导致了以性能为导向的方法，因为其根本目标是生产出满足特定性能目标的产品。 

4  The theoretical approach leads to a simulation-oriented methodology because the underlying goal is to expand our understanding of intelligence and thus the emphasis is on the underlying process rather than the exterior performance.

理论方法导致了一种以模拟为导向的方法，因为其根本目标是扩展我们对智能的理解，因此重点在于底层过程而不是外部表现。

5  As an example, consider the fields of natural language processing and linguistics. These fields are closely related and benefit from research in each other, yet the underlying goals are different. Linguists are interested in learning how humans process language and thus tend toward more theoretical pursuits.   

以自然语言处理和语言学领域为例。这些领域密切相关，相互研究有益，但其根本目标不同。语言学家对了解人类如何处理语言感兴趣，因此更倾向于理论研究

6 Researchers in the field of natural language processing are interested in developing machines that can manipulate natural language and therefore lean in the engineering direction.

7 Researchers in the field of natural language processing are interested in developing machines that can manipulate natural language and therefore lean in the engineering direction.

自然语言处理领域的研究人员对开发能够操纵自然语言的机器感兴趣，因此倾向于工程方向。

8 Thus, linguists operate in simulationoriented mode—building systems whose goals are to test theories. In contrast,researchers in natural language processing operate in performance-oriented
mode—building systems to perform tasks.  

因此，语言学家以模拟为导向的模式工作——构建系统的目的是测试理论。相比之下，自然语言处理研究人员则以性能为导向的模式工作——构建系统来执行任务。

9 Systems produced in this latter mode (such as document translators and systems by which machines respond to verbal commands) rely heavily on knowledge gained by linguists but often apply “shortcuts” that happen to work in the restricted environment of the particular system.  

在后一种模式下产生的系统（例如文档翻译器和机器响应口头命令的系统）在很大程度上依赖于语言学家获得的知识，但通常会在特定系统的有限环境中使用“捷径”。

10 As an elementary example, consider the task of developing a shell for an operating system that receives instructions from the outside world through verbal English commands. In this case, the shell (an agent) does not need to worry about the entire English language.  

作为一个基本的例子，考虑为一个操作系统开发一个shell的任务，该操作系统通过口头英语命令接收来自外部世界的指令。在这种情况下，shell(代理)不需要担心整个英语语言。

11  More precisely, the shell does not need to distinguish between the various meanings of the word copy. (Is it a noun or a verb? Should it carry the connotation of plagiarism?) Instead, the shell needs merely to distinguish the word copy from other commands such as rename and delete.

更确切地说，shell 不需要区分 copy 这个词的各种含义。（它是名词还是动词？它是否应该带有抄袭的含义？）相反，shell 只需要将 copy 一词与其他命令（如 rename 和 delete）区分开来。

12  Thus the shell could perform its task just by matching its inputs to predetermined audio patterns. The performance of such a system may be satisfactory to an engineer but the way it is obtained would not be aesthetically pleasing to a theoretician. 

因此，shell只需将其输入与预定的音频模式相匹配即可执行其任务。这种系统的性能可能令工程师满意，但其实现方式对理论家来说并不美观。

三、Turing Test

1. 名词解释  

1 Turing Test

The Turing test measures whether a machine has the ability to exhibit intelligent behaviors that are indistinguishable from humans. It was created by Alan Turing and involves a human evaluator who communicates with machines through a typewriter system without seeing each other. If this machine is often mistaken for a human, it can pass the test.

2.翻译 

1  In the past the Turing test (proposed by Alan Turing in 1950) has served as a benchmark in measuring progress in the field of artificial intelligence. Today the significance of the Turing test has faded although it remains an important part of the artificial intelligence folklore. 

过去，图灵测试（由艾伦·图灵于 1950 年提出）曾作为衡量人工智能领域进展的标杆。如今，图灵测试的意义已日渐淡化，尽管它仍然是人工智能传说中的重要组成部分。

2  Turing’s proposal was to allow a human, whom we call the interrogator, to communicate with a test subject by means of a typewriter system without being told whether the test subject was a human or a machine. In this environment, a machine would be declared to behave intelligently if the interrogator was not able to distinguish it from a human.

图灵的提议是允许一个人（我们称之为询问者）通过打字机系统与测试对象进行交流，而无需告知测试对象是人还是机器。在这种环境下，如果询问者无法将机器与人类区分开来，则机器将被宣布为具有智能行为。 

3  Turing predicted that by the year 2000 machines would have a 30 percent chance of passing a five-minute Turing test—a conjecture that turned out to be surprisingly accurate.

图灵预测，到 2000 年，机器通过五分钟图灵测试的概率将为 30%——这一猜想最终被证明
出人意料地准确

4 One reason that the Turing test is no longer considered to be a meaningful measure of intelligence is that an eerie appearance of intelligence can be produced with relative ease.

图灵测试不再被视为衡量智力的有效方法，原因之一是，智力的怪异表象可以相对轻松地产生。

5  A well-known example arose as a result of the program DOCTOR (a version of the more general system called ELIZA) developed by Joseph Weizenbaum in the mid-1960s. 

一个著名的例子是 Joseph Weizenbaum 在 20 世纪 60 年代中期开发的 DOCTOR 程序（更通用的系统 ELIZA 的一个版本）

6  This interactive program was designed to project the image of a Rogerian analyst conducting a psychological interview; the computer played the role of the analyst while the user played the patient.

这个互动程序旨在投射罗杰斯分析师进行心理访谈的形象；计算机扮演分析师的角色，而用户扮演患者 

7 Internally, all that DOCTOR did was restructure the statements made by the patient according to some well-defined rules and direct them back to the patient. For example, in response to the statement “I am tired today,” DOCTOR might have replied with “Why do you think you’re tired today?” If DOCTOR was unable to recognize the sentence structure, it merely responded with something like “Go on” or “That’s very interesting.” 

在内部，DOCTOR 所做的只是根据一些明确定义的规则重构患者的陈述，并将其反馈给患者。例如，在回应“我今天很累”这句话时，DOCTOR 可能会回答“你为什么认为你今天很累？”如果 DOCTOR 无法识别句子结构，它只会回答“继续”或“这很有趣”之类的话。

8 Weizenbaum’s purpose in developing DOCTOR dealt with the study of natural language communication. The subject of psychotherapy merely provided an environment in which the program could “communicate.” 

魏森鲍姆开发 DOCTOR 的目的是研究自然语言交流。心理治疗这一主题只是为程序提供了一个可以“交流”的环境。

9  To Weizenbaum’s dismay, however, several psychologists proposed using the program for actual psychotherapy. (The Rogerian thesis is that the patient, not the analyst, should lead the discussion during the therapeutic session, and thus, they argued, a computer could possibly conduct a discussion as well as a therapist could.) 

然而，令魏森鲍姆失望的是，一些心理学家建议将该程序用于实际的心理治疗。（罗杰斯的论点是，在治疗过程中，应该由患者而不是分析师来主导讨论，因此，他们认为，计算机可以像治疗师一样进行讨论。）

10  Moreover, DOCTOR projected the image of comprehension so strongly that many who “communicated” with it became subservient to the machine’s question-and-answer dialogue. In a sense, DOCTOR passed the Turing test. The result was that ethical, as well as technical, issues were raised, and Weizenbaum became an advocate for maintaining human dignity in a world of advancing technology.  

此外，DOCTOR 如此强烈地投射出理解的形象，以至于许多与它“交流”的人都屈从于机器的问答对话。从某种意义上说，DOCTOR 通过了图灵测试。结果是提出了道德和技术问题，而魏森鲍姆则成为了在技术进步的世界中维护人类尊严的倡导者。

11  More recent examples of Turing test “successes” include Internet viruses that carry on “intelligent” dialogs with a human victim in order to trick the human into dropping his or her malware guard. Moreover, phenomena similar to Turing tests occur in the context of computer games such as chess-playing programs.

图灵测试“成功”的近期例子包括互联网病毒，它与人类受害者进行“智能”对话，以诱骗人类放下恶意软件防御。此外，类似于图灵测试的现象也发生在计算机游戏（如国际象棋程序）的背景下。  

12  Although these programs select moves merely by applying brute-force techniques (similar to those we will discuss in Section 11.3), humans competing against the computer often experience the sensation that the machine possesses creativity and even a personality.   

尽管这些程序只是通过应用蛮力技术来选择动作（类似于我们将在第 11.3 节中讨论的技术），但与计算机竞争的人类经常会体验到机器具有创造力甚至个性的感觉。 

13  Similar sensations occur in robotics where machines have been built with physical attributes that project intelligent characteristics. Examples include toy robot dogs that project adorable personalities merely by tilting their heads or lifting their ears in response to a sound.  

类似的感觉也出现在机器人领域，机器人的物理属性可以投射出智能特征。例如，玩具机器狗只需倾斜头部或抬起耳朵来响应声音，就能展现出可爱的个性。

四、Perception

1. 名词解释  

1 image processing

image processing,which refers to identifying characteristics of the image 

2 image analysis 

image analysis, which refers to the process of understanding what these characteristics mean.

2.翻译  

1  To respond intelligently to the input from its sensors, an agent must be able to understand that input. That is, the agent must be able to perceive. In this section we explore two areas of research in perception that have proven to be especially challenging—understanding images and language.

为了智能地响应来自传感器的输入，代理必须能够
理解该输入。也就是说，代理必须能够感知。在本节中，我们将探讨感知领域的两个研究领域，这两个领域已被证明特别具有挑战性——理解图像和语言。 

2  Let us consider the problems posed by the puzzle-solving machine introduced in the previous section. The opening and closing of the gripper on the machine presents no serious obstacle, and the ability to detect the presence of the puzzle in the gripper during this process is straightforward because our application requires very little precision.  

让我们考虑一下上一节介绍的解谜机所带来的问题。机器上夹持器的打开和关闭不会造成严重障碍，并且在此过程中检测夹持器中拼图的存在的能力很简单，因为我们的应用程序要求的精度非常低。

3  Even the problem of focusing the camera on the puzzle can be handled simply by designing the gripper to position the puzzle at a particular predetermined position for viewing. Consequently, the first intelligent behavior required by the machine is the extraction of information through a visual medium.  

即使是将镜头对准谜题的问题也可以通过设计抓手将谜题放置在特定的预定位置来解决。因此，机器需要的第一个智能行为是通过视觉媒介提取信息。 

4  It is important to realize that the problem faced by our machine when looking at the puzzle is not that of merely producing and storing an image. Technology has been able to do this for years as in the case of traditional photography and television systems.  

重要的是要意识到，我们的机器在看拼图时面临的问题不仅仅是生成和存储图像。技术多年来一直能够做到这一点，就像传统摄影和电视系统的情况一样。  

5  Instead, the problem is to understand the image in order to extract the current status of the puzzle (and perhaps later to monitor the movement of the tiles). 

相反，问题在于理解图像，以便提取谜题的当前状态(也许以后还要监视贴图的移动)。  

6  In the case of our puzzle-solving machine, the possible interpretations of the puzzle image are relatively limited. We can assume that what appears is always an image containing the digits 1 through 8 in a well-organized pattern.  

以我们的解谜机为例，谜题图像的可能解释是相对有限的。我们可以假设出现的总是一个包含数字1到8的图像，其模式组织良好。 

7  The problem is merely to extract the arrangement of these digits. For this, we imagine that the picture of the puzzle has been encoded in terms of bits in the computer’s memory, with each bit representing the brightness level of a particular pixel.

问题仅仅是要找出这些数字的排列。为此，我们想象这个谜题的图片已经在计算机内存中以比特的形式进行了编码，每个比特代表一个特定像素的亮度水平。  

8  Assuming a uniform size of the image (the machine holds the puzzle at a predetermined location in front of the camera), our machine can detect which tile is in which position by comparing the different sections of the picture to prerecorded templates consisting of the bit patterns produced by the individual digits used in the puzzle. As matches are found, the condition of the puzzle is revealed.  

假设图像的大小是一致的(机器将拼图放在相机前面的一个预先确定的位置)，我们的机器可以通过将图片的不同部分与预先录制的模板进行比较来检测哪个瓷砖在哪个位置，这些模板由拼图中使用的单个数字产生的位模式组成。当找到匹配时，谜题的情况就会显露出来。  

9  This technique of recognizing images is one method used in optical character readers. It has the drawback, however, of requiring a certain degree of uniformity for the style, size, and orientation of the symbols being read.   

这种识别图像的技术是光学字符阅读器中使用的一种方法。但它的缺点是要求所读取符号的样式、大小和方向具有一定的统一性。 

10  In particular, the bit pattern produced by a physically large character does not match the template for a smaller version of the same symbol, even though the shapes are the same. Moreover, you can imagine how the problems increase in difficulty when trying to process handwritten material. 

尤其是，一个物理上较大的字符产生的位模式与相同符号的较小版本的模板不匹配，即使形状相同。此外，你可以想象在尝试处理手写材料时，问题的难度会增加多少。  

11  Another approach to the problem of character recognition is based on matching the geometric characteristics rather than the exact appearance of the symbols. In such cases the digit 1 might be characterized as a single vertical line, 2 might be an opened curved line joined with a horizontal straight line across the bottom, and so on. 

字符识别问题的另一种方法是基于匹配几何特征而不是符号的确切外观。在这种情况下，数字1可能被表征为一条垂直线，2可能是一条打开的曲线，并在底部连接一条水平直线，以此类推。

12 This method of recognizing symbols involves two steps: the first is to extract the features from the image being processed, and the second is to compare the features to those of known symbols.As with the template-matching approach, this technique for recognizing characters is not foolproof.

这种识别符号的方法分为两个步骤:第一步是从待处理图像中提取特征，第二步是将特征与已知符号的特征进行比较。与模板匹配方法一样，这种识别字符的技术也不是万无一失的。 

13 For instance, minor errors in the image can produce a set of entirely different geometric features, as in the case of distinguishing between an O and a C or, in the case of the eight-puzzle, a 3 and an 8. 

例如，图像中的微小错误可能会产生一组完全不同的几何特征，例如在区分O和C的情况下，或者在8字谜的情况下，区分3和8。

14  We are fortunate in our puzzle application because we do not need to understand images of general three-dimensional scenes. Consider, for example, the advantage we have by being assured that the shapes to be recognized (the digits 1 through 8) are isolated in different parts of the picture rather than appearing as overlapping images, as is common in more general settings. 
在我们的拼图应用程序中，我们很幸运，因为我们不需要理解一般三维场景的图像。例如，考虑一下我们的优势，我们确信要识别的形状(数字1到8)在图片的不同部分是孤立的，而不是像在更一般的设置中那样出现重叠的图像。 

15  In a general photograph, for instance, one is faced not only with the problem of recognizing an object from different angles but also with the fact that some portions of the object might be hidden from view. 

例如，在一张普通照片中，人们不仅面临着从不同角度识别物体的问题，而且还面临着物体的某些部分可能隐藏在视线之外的事实。

16 The task of understanding general images is usually approached as a twostep process: (1) image processing, which refers to identifying characteristics of the image, and (2) image analysis, which refers to the process of understanding what these characteristics mean. 

理解一般图像的任务通常被视为两个步骤的过程:(1)图像处理，这是指识别图像的特征;(2)图像分析，这是指理解这些特征意味着什么的过程 

17 We have already observed this dichotomy in the context of recognizing symbols by means of their geometric features.  In that situation, we found image processing represented by the process of identifying the geometric features found in the image and image analysis represented by the process of identifying the meaning of those features.

我们已经在通过几何特征识别符号的背景下观察到了这种二分法。在这种情况下，我们发现以识别图像中发现的几何特征的过程为代表的图像处理和以识别这些特征的含义为代表的图像分析。 

18 Image processing entails numerous topics. One is edge enhancement, which is the process of applying mathematical techniques to clarify the boundaries between regions in an image. 

图像处理涉及许多主题。一种是边缘增强，这是应用数学技术来澄清图像中区域之间边界的过程。 

19  In a sense, edge enhancement is an attempt to convert a photograph into a line drawing. Another activity in image analysis is known as region finding. 

从某种意义上说，边缘增强是将照片转换成线条图的一种尝试。图像分析中的另一项活动被称为区域查找。

20  This is the process of identifying those areas in an image that have common properties such as brightness, color, or texture. Such a region probably represents a section of the image that belongs to a single object. (It is the ability to recognize regions that allows computers to add color to old-fashioned black and white motion pictures.) 

这是识别图像中具有共同属性(如亮度、颜色或纹理)的区域的过程。这样的区域可能代表属于单个对象的图像的一部分。(正是这种识别区域的能力，使计算机能够为老式的黑白电影添加色彩。)

21  Still another activity within the scope of image processing is smoothing, which is the process of removing flaws in the image. Smoothing keeps errors in the image from confusing the other image-processing steps, but too much smoothing can cause the loss of important information as well. 

图像处理范围内的另一项活动是平滑，即去除图像中缺陷的过程。平滑可以防止图像中的错误混淆其他图像处理步骤，但过多的平滑也会导致重要信息的丢失。 

22  Smoothing, edge enhancement, and region finding are all steps toward identifying the various components in an image. Image analysis is the process of determining what these components represent and ultimately what the image means.  

平滑、边缘增强和区域查找都是识别图像中各种成分的步骤。图像分析是确定这些组件代表什么并最终确定图像含义的过程。  

23  Here one faces such problems as recognizing partially obstructed objects from different perspectives.  One approach to image analysis is to start with an assumption about what the image might be and then try to associate the components in the image with the objects whose presence is conjectured.

在这里，人们面临着从不同角度识别部分被遮挡物体的问题。图像分析的一种方法是从对图像可能是什么的假设开始，然后尝试将图像中的组件与推测存在的对象联系起来  

24  This appears to be an approach applied by humans.  For instance, we sometimes find it hard to recognize an unexpected object in a setting in which our vision is blurred, but once we have a clue to what the object might be, we can easily identify it.

这似乎是人类应用的一种方法。例如，我们有时会发现，在视觉模糊的环境中，很难识别出一个意想不到的物体，但一旦我们有了这个物体可能是什么的线索，我们就能很容易地识别出来。  

25  The problems associated with general image analysis are enormous, and much research in the area remains to be done. Indeed, image analysis is one of the fields that demonstrates how tasks that are performed quickly and apparently easily by the human mind continue to challenge the capabilities of machines.  

与一般图像分析相关的问题是巨大的，在该领域还有许多研究有待完成。事实上，图像分析是一个领域，它展示了人类大脑如何快速、轻松地完成任务，并继续挑战机器的能力。    

五、Language Processing

1. 名词解释  

1 semantic analysis 

semantic analysis is charged with the task of identifying the semantic role of each word in the statement. Semantic analysis seeks to identify such things as the action described, the agent of that action and the object of the action.

2 contextual analysis

It is at this level that the context of the sentence is brought into the understanding process.

2.翻译  

1  Another perception problem that has proven challenging is that of understanding language. The success obtained in translating formal high-level programming languages into machine language (Section 6.4) led early researchers to believe that the ability to program computers to understand natural language was only a few years away. 

另一个被证明具有挑战性的感知问题是理解语言。在将正式的高级编程语言翻译成机器语言方面取得的成功(第6.4节)使早期的研究人员相信，通过编程让计算机理解自然语言的能力只需要几年的时间。  

2  Indeed, the ability to translate programs gives the illusion that the machine actually understands the language being translated. (Recall from Section 6.1 the story told by Grace Hopper about managers who thought she was teaching computers to understand German.) 

事实上，翻译程序的能力给人一种错觉，即机器实际上理解被翻译的语言。(回想一下第6.1节中格蕾丝·霍珀(Grace Hopper)讲述的一个故事，她说经理们认为她是在教计算机理解德语。) 

3  What these researchers failed to understand was the depth to which formal programming languages differ from natural languages such as English, German, and Latin. Programming languages are constructed from well-designed primitives so that each statement has only one grammatical structure and only one meaning.  

这些研究人员未能理解的是，形式编程语言与自然语言(如英语、德语和拉丁语)的深度不同。编程语言是由设计良好的原语构成的，因此每个语句只有一种语法结构和一种含义。 

4  In contrast, a statement in a natural language can have multiple meanings depending on its context or even the manner in which it is communicated.  Thus, to understand natural language, humans rely heavily on additional knowledge.

相比之下，自然语言中的语句可以有多种含义，这取决于它的上下文甚至是交流的方式。因此，为了理解自然语言，人类严重依赖于额外的知识。  

5  For example, the sentences   Norman Rockwell painted people. and Cinderella had a ball.
have multiple meanings that cannot be distinguished by parsing or translating eachword independently.    Instead, to understand these sentences requires the ability to comprehend the context in which the statement is made.  In other instances the true meaning of a sentence is not the same as its literal translation.  

例如，这些句子
诺曼·洛克威尔画人物。
和
灰姑娘有一个舞会。
有多个含义，不能通过单独解析或翻译每个单词来区分。相反，要理解这些句子，需要有能力理解句子所处的语境。在其他情况下，句子的真正含义与字面翻译并不相同。 

6  For example, Do you know what time it is? often means “Please tell me what time it is,” or if the speaker has been waiting for a long time, it might mean “You are very late.”
To unravel the meaning of a statement in a natural language therefore requires several levels of analysis.  

例如, 你知道现在几点了吗?

通常的意思是“请告诉我现在几点了”，或者如果说话人等了很长时间，它的意思可能是“你迟到了”。因此，要在自然语言中揭示语句的含义，需要几个层次的分析。 

7  The first of these is syntactic analysis, whose major component is parsing. It is here that the subject of the sentence Mary gave John a birthday card.
is recognized as Mary while the subject of John got a birthday card from Mary. is found to be John.  

第一种是语法分析，其主要组成部分是解析。在这里，句子的主语玛丽给了约翰一张生日卡。
主人公约翰被认出是玛丽，而主人公约翰收到了玛丽寄来的生日贺卡。被发现是约翰。  

8  Another level of analysis is called semantic analysis. In contrast to the parsing process, which merely identifies the grammatical role of each word, semantic analysis is charged with the task of identifying the semantic role of each word in the statement.  

另一个层次的分析称为语义分析。解析过程仅仅识别每个单词的语法作用，而语义分析的任务是识别语句中每个单词的语义作用。

9  Semantic analysis seeks to identify such things as the action described, the agent of that action (which might or might not be the subject of the sentence), and the object of the action. It is through semantic analysis that the sentences “Mary gave John a birthday card” and “John got a birthday card from Mary” would be recognized as saying the same thing.   

语义分析试图识别诸如所描述的动作、该动作的代理(可能是也可能不是句子的主语)和动作的对象等内容。通过语义分析，“玛丽给了约翰一张生日卡片”和“约翰从玛丽那里得到了一张生日卡片”这两个句子可以被认为是同一件事。

10 A third level of analysis is contextual analysis. It is at this level that the context of the sentence is brought into the understanding process. For example, it is easy to identify the grammatical role of each word in the sentence The bat fell to the ground.  

第三个层次的分析是上下文分析。正是在这个层次上，句子的语境被带入了理解过程。例如，很容易识别句子中每个单词的语法作用  bat fell to the ground。 

11  We can even perform semantic analysis by identifying the action involved as falling, the agent as bat, and so on.  But it is not until we consider the context of the statement that the meaning of the statement becomes clear.  In particular, it has a different meaning in the context of a baseball game than it does in the context of cave exploration.
我们甚至可以通过将所涉及的动作识别为坠落、代理为蝙蝠等来执行语义分析。但是，直到我们考虑这句话的上下文，这句话的意义才变得清楚。特别是，它在棒球比赛中的含义与在洞穴探险中的含义不同。  

12  Moreover, it is at the contextual level that the true meaning of the question “Do you know what time it is?” would finally be revealed. We should note that the various levels of analysis—syntactic, semantic, and contextual—are not necessarily independent. 

此外，只有在语境层面，“你知道现在几点了吗？”这个问题的真正含义才会最终显露出来。我们应该注意到，句法、语义和语境等各个分析层面并不一定是独立的。