Answers are in blue.

Computer Science Illuminated, Seventh Edition Nell Dale, PhD; John Lewis, PhD

CHAPTER 13 EXERCISES AND ANSWERS

?

For Exercises 1–5, match the type of ambiguity with an example.

1. Lexical詞匯歧義
2. Referential指代歧義
3. Syntactic句法
   1. “Stand up for your flag.”
      1. Does “stand up” mean support or rise to your feet?
   2. “Go down the street on the left.”

C. Does this mean to go left down the street or go down the street that is on the left?

1. 1. “He drove the car over the lawn mower, but it wasn’t hurt.”

B. What wasn’t hurt, the car or the lawn mower?

1. 1. “I saw the movie flying to Houston.” C
   2. “Mary and Kay were playing until she came inside.” B

For Exercises 6–21, mark the answers true and false as follows:

1. 1. 1. True
      2. False
   2. A computer does some tasks much better than a human being.

A

1. 1. A human being does some tasks much better than a computer.

A

1. 1. A computer system that can pass the Turing test is consid- ered to be intelligent.

A

1. 1. Some AI researchers don’t think we can achieve true artificial intelligence until a computer processes information in the same way the human mind does.

A

1. 1. A semantic network is used to model relationships. A
   2. 語義網絡用于建模關系。
   3. If information is stored in a semantic network, it is easy to answer questions about it.

B (it depends on how the network is structured)

如果信息存儲在語義網絡中，回答相關問題會很容易。

B（這取決于網絡的結構）

1. 1. A computer has never beaten a human at chess in master- level play.

B

?

1. 1. An inference engine is part of a rule-based expert system. A
   2. A biological neuron accepts a single input signal and pro- duces multiple output signals.

B

生物神經元多個輸入一個輸出

1. 1. Each element in an artificial neural net is affected by a numeric weight.

A

1. 1. Voice synthesis is the most difficult part of natural language processing.

B語音合成不是最苦難的

1. 1. Each human has a unique voiceprint that can be used to train voice recognition systems.

A

1. 1. The word “light” can be interpreted in many ways by a computer.

A

1. 1. Syntactic ambiguity is no longer a problem for natural lan- guage comprehension.

B

1. 1. A robot may follow the sense–plan–act approach to control its movements.

A

1. 1. Isaac Asimov created three fundamental laws of robotics. A
   2. 6. 計算機在某些任務上比人類做得更好。
   3. A
   4. 7. 人類在某些任務上比計算機做得更好。
   5. A
   6. 8. 能通過圖靈測試的計算機系統被認為是智能的。
   7. A
   8. 9. 一些人工智能研究人員認為，只有當計算機以與人類思維相同的方式處理信息時，我們才能實現真正的人工智能。
   9. A
   10. 10. 語義網絡用于建模關系。
   11. A
   12. 11. 如果信息存儲在語義網絡中，回答相關問題會很容易。
   13. B（這取決于網絡的結構）
   14. 12. 計算機在大師級別的國際象棋比賽中從未戰勝過人類。
   15. B
   17. 13. 推理引擎是基于規則的專家系統的一部分。
   18. A
   19. 14. 生物神經元接受單一輸入信號并產生多個輸出信號。
   20. B
   21. 15. 人工神經網絡中的每個元素都受到數值權重的影響。
   22. A
   23. 16. 語音合成是自然語言處理中最困難的部分。
   24. B
   25. 17. 每個人都有一個獨特的聲紋，可用于訓練語音識別系統。
   26. A
   27. 18. 計算機可以以多種方式解釋單詞“light”。
   28. A
   29. 19. 對于自然語言理解，句法歧義不再是問題。
   30. B
   31. 20. 機器人可能遵循感知-計劃-執行方法來控制其動作。
   32. A
   33. 21. 艾薩克·阿西莫夫創立了機器人學的三個基本定律。
   34. A

For Exercises 22–30, match the task with which (human or computer) can solve it more easily.

1. 1. 1. Computer
      2. Human
   2. Identify a dog in a picture. B
   3. Add a column of 100 four-digit numbers. A
   4. Interpret a poem. B
   5. Match a finger print. A
   6. Paint a landscape. B
   7. Carry on an intelligent conversation. B

對于練習22-30，將任務與更容易解決它的一方（人類或計算機）進行匹配。

A. 計算機

B. 人類

56. 識別圖片中的狗。 B

57. 添加100個四位數的列。 A

58. 解釋一首詩。 B

59. 匹配指紋。 A

60. 繪制一幅風景畫。 B

61. 進行智能對話。 B

1. 1. Learn to speak. B
   2. Judge guilt or innocence. B
   3. Give affection. B

Exercises 31–76 are problems or short-answer questions.

1. 1. What is the Turing test?

The Turing test is a test devised by Alan Turing to answer the question “How can we know we’ve succeeded in creating a machine that can think?” The test is based on whether a computer could fool a human into believing that the com- puter is another human being.

靈測試是由艾倫·圖靈設計的一種測試，旨在回答“我們如何知道我們成功地創建了一臺能夠思考的機器？”的問題。該測試基于計算機是否能夠欺騙人類，使其相信計算機是另一名人類。

1. 1. How is the Turing test organized and administered?

A human interrogator sits in a room and uses a computer terminal to communicate with two respondents. The interro- gator knows that one respondent is human and the other is a computer. After conversing with both the human and the computer, the interrogator must decide which respondent is the computer. If the computer could fool enough interroga- tors, then it must be considered intelligent.

一位人類詢問者坐在一個房間里，使用計算機終端與兩個被詢問者進行通信。詢問者知道其中一個被詢問者是人類，另一個是計算機。在與人類和計算機都進行對話之后，詢問者必須決定哪個被詢問者是計算機。如果計算機能夠欺騙足夠多的詢問者，那么它就被認為是具有智能的。

1. 1. What is weak equivalence and how does it apply to the Turing test?

Weak equivalence is the equality of two systems based on their results. The Turing test shows weak equivalence.

弱等價性是基于兩個系統的結果的相等性。圖靈測試展示了弱等價性。

1. 1. What is strong equivalence?

Strong equivalence is the quality of two systems based on their results and the process by which they arrive at those results.

強等價性是基于兩個系統的結果以及它們達到這些結果的過程的質量

1. 1. What is the Loebner prize?

The Loebner prize is the first formal instantiation of the Tur- ing test. It has been held annually since 1991.

洛布納獎是圖靈測試的第一個正式實例。自1991年以來，它每年都會舉行

1. 1. Name and describe briefly five issues in the world of AI cov- ered in this chapter.

Knowledge representation: The techniques used to rep- resent knowledge so that a computer system can use it in problem solving.

Expert systems: Computer systems that embody the knowl- edge of human experts.

Neural networks: Computer systems that mimic the process- ing of the human brain.

Natural-language processing: Computer systems that pro- cess the language that humans use to communicate.

Robotics: The study of mobile robots that use AI techniques to interact with their environments.

知識表示：用于表示知識以便計算機系統在問題解決中使用的技術。

專家系統：具有人類專家知識的計算機系統。

神經網絡：模擬人腦處理過程的計算機系統。

自然語言處理：處理人類用于交流的語言的計算機系統。

機器人學：研究使用人工智能技術與環境互動的移動機器人。

1. 1. Name and define two knowledge representation techniques. Semantic networks: A technique that represents the relation- ships among objects.

Search trees: A structure that represents alternatives in adversarial situations such as games.

語義網絡：一種表示對象之間關系的技術。

搜索樹：表示在對抗性情境中的替代方案的結構。

?

1. 3. What data structure defined in Chapter 8 is used to represent a semantic network?

A graph is used to represent a semantic network. The nodes in the graph represent objects and the arrows (arcs) repre- sent relationships.

1. 1. Create a semantic network for the relationships among your family members. List five questions that your semantic net could easily be used to answer and five questions that would be more of a challenge to answer.

?

Easy questions to answer given this organization: Who are John’s children?

What is the gender of Kayla? How old are Sharon’s children?

How many female children does John have?

Does Sharon have any children older than 5 years of age?

More challenging questions to answer given this organization: Who are Kayla’s parents?

Who are Justin’s siblings?

How many female children are there? Who is the mother of John’s children? Does John have any step-children?

1. 1. Create a semantic network that captures the information in a small section of a newspaper article.

This is an activity for which no answer is appropriate.

1. 1. What object-oriented properties do semantic networks borrow?

Semantic networks borrow inheritance and instantiation. The inheritance is expressed in the “is-a” relationship, and instanti- ation is expressed when an object is related to something that describes it.

語義網絡借用了繼承和實例化。繼承表達了“是一個”關系，而實例化是指當一個對象與描述它的東西相關聯時。

1. 1. What is a search tree?

A search tree is a structure that represents all possible moves for both players in a two-person game.

搜索樹是一種表示兩人游戲中雙方所有可能走法的結構。

1. 1. What are trees for complex games like chess to large?

A search tree contains all possible moves from the first posi- tion, all possible moves from each of the moves from the first position, , all possible moves from all possible moves at

the level above. Thus the trees are very large for complex games like chess.

對于復雜游戲如國際象棋，樹的結構會變得很大。

搜索樹包含了從初始位置開始的所有可能移動，以及每個從初始位置開始的移動的所有可能移動，以此類推，一直到上一級的所有可能移動。因此，對于復雜游戲如國際象棋，樹的結構會非常龐大。

1. 1. Distinguish between depth-first searching and breadth-first searching.

Depth-first searching begins at the top level (root) and con- tinues going deeper and deeper into the tree until the search has reached a leaf node, at which time the search moves back up one level and starts down again. A breadth-first search begins at the top level, then searches every node on the next lower level, then searches every node at the next lower level, until it has searched every node on every level.

深度優先搜索從頂層（根）開始，不斷深入樹中，直到搜索到達葉節點，此時搜索會回溯到上一層并重新開始。而廣度優先搜索則從頂層開始，然后搜索下一層的每個節點，接著搜索下一層的每個節點，直到搜索完每個層的每個節點。

1. 1. What does it mean to prune a tree?

Pruning a tree means to eliminate some branches from searching.

剪枝是指從搜索中消除一些分支。

1. 1. Distinguish between knowledge-based systems and expert systems.

A knowledge-based system is a software system that uses a specific set of information from which it extracts and pro- cesses particular pieces. An expert system is sometime used as a synonym, but it also carries with it the idea of modeling the expertise of a professional in that particular field.

基于知識系統是一種軟件系統，它使用特定的信息集，從中提取和處理特定的信息片段。專家系統有時被用作同義詞，但它還帶有模擬該領域專業人士的專業知識的概念。

1. 1. Distinguish be rule-based systems and inference engines.

A rule-based system is a software system that uses a set of rules to guide its processing. An inference engine is the soft- ware system that processes the rules.

基于規則系統是一種軟件系統，它使用一組規則來引導其處理。推理引擎是處理規則的軟件系統。

1. 1. What is an example of a human expert system?

A doctor is an example of a human expert system. The doc- tor asks questions and runs tests based on his knowledge and experience.

醫生是人類專家系統的一個例子。醫生根據其知識和經驗詢問問題并進行測試。

1. 1. What do we call a knowledge-based system that models the expertise of professionals in the field?

An expert system.

專家系統被稱為基于規則的系統，因為它使用一組規則來引導其處理

1. 1. Why is an expert system called a rule-based system?

An expert system is called a rule-based system because it uses a set of rules to guide its processing.

1. 1. What is the part of the software in an expert system that determines how the rules are followed and what conclusions can be drawn?

An inference engine.

推理引擎。

1. 1. How are the rules expressed in an expert system?

The rules are expressed as selection statements (if statements).

規則被表達為選擇語句（if 語句）。

1. 1. What are the advantages of an expert system?

An expert system is goal oriented; it doesn’t focus on abstract or theoretical information. It is efficient; it records previous responses and doesn’t ask irrelevant questions. An expert system can provide useful guidance even if it can’t provide the answer to a specific question.

專家系統是目標導向的；它不關注抽象或理論性的信息。它高效；它記錄先前的響應并不問無關的問題。即使不能回答特定問題，專家系統仍然可以提供有用的指導。

1. 1. What is a single cell that conducts a chemically based elec- tronic signal?

A neuron.

1. 1. What do a series of connected neurons form? A pathway in the brain.

?

1. 2. Upon what does the signal along a particular pathway depend? The signals depend on the state of the neurons through which the signal passes.
   3. 信號沿特定通路的傳播取決于什么？信號取決于沿途神經元的狀態。
   4. 生物神經元中的多輸入觸角是什么？樹突。
   6. What are the multiple input tentacles in a biological neuron? Dendrites
   7. 樹突
   8. What is the primary output tentacle in a biological neuron? An axon.
   9. 生物神經元中的主要輸出觸角是什么？軸突。
   10. 軸突
   11. From where do dendrites of one neuron pick up the signals from other neurons to form a network?

The dendrites of one neuron pick up the signals from the axons of other neurons to forma neural network.

一個神經元的樹突從其他神經元的軸突接收信號，形成神經網絡。

1. 1. What is the gap between an axon and a dendrite? A synapse.
   2. 軸突和樹突之間有什么間隙？突觸。
   3. What tempers the strength of a synapse?

The chemical composition of a synapse tempers the strength of its input signal.

什么調節了突觸的強度？突觸的化學成分調節了其輸入信號的強度。

1. 1. What is the role of a synapse?

The role of a synapse is to weight the input signal.

突觸的作用是加權輸入信號。

1. 1. How is a synapse modeled in an artificial neural network?

A synapse is represented by a weight assigned to each input signal.

如何在人工神經網絡中建模突觸？突觸由分配給每個輸入信號的權重表示。

1. 1. What is an effective weight in an artificial neuron?

An effective weight is the sum of the weights multiplied by the corresponding input values.

工神經元中的有效權重是什么？有效權重是權重乘以相應輸入值的總和。

1. 1. How is the output value from an artificial neuron calculated? Each neuron has a numeric threshold value. If the effective weight is greater than the threshold, a 1 is output; otherwise, a 0 is output.
   2. 人工神經元的輸出值如何計算？每個神經元都有一個數值閾值。如果有效權重大于閾值，則輸出為1；否則，輸出為0。
   3. If the processing element in an artificial neural net accepted five input signals with values of 0, 0, 1, 1, and 0 and corre- sponding weights of 5, -2, 3, 3, and 6, what is the output if the threshold is 5?

1

如果人工神經網絡中的處理元素接受了值為0、0、1、1和0的五個輸入信號，相應的權重為5、?2、3、3和6，如果閾值為7，輸出是什么？

0

1. 1. If the processing element in an artificial neural net accepted five input signals with values of 0, 0, 1, 1, and 0 and corre- sponding weights of 5, -2, 3, 3, and 6, what is the output if the threshold is 7?

0

1. 1. What is a phoneme?

A phoneme is a fundamental sound in a language.

什么是音素？

音素是語言中的基本音

1. 1. Describe the two distinct ways that voice synthesis can be accomplished.

In dynamic voice generation, the set of phonemes for a lan- guage are generated. A computer examines the letters that make up a word and produce the sequence of sounds using the language’s phonemes.

In recorded speech, human speech is recorded. A computer chooses the correct word from its file of recorded words.

Dynamic voice generation can make an attempt to pro- nounce any word, but recorded speech can only pronounce words that have been prerecorded.

描述實現語音合成的兩種明顯方式。

在動態語音生成中，生成語言的音素集。計算機檢查組成一個單詞的字母，并使用語言的音素生成聲音序列。

在記錄語音中，記錄人類的語音。計算機從其已記錄的單詞文件中選擇正確的單詞。

動態語音生成可以嘗試發音任何單詞，但記錄的語音只能發音已經預先記錄的單詞。

1. 1. Which issues affect the ability to recognize the words spoken by a human voice?

Accents, regional dialects, voice pitch, homonyms, and the clarity of a person’s speech.

口音、地方方言、語調、同音異義詞以及個人言語的清晰度。