Vr input devices

Task 12

Reading

Read quickly through the text below. Does the text contradict in any way what you already know about VR systems? If so, what is the contradiction and how can it be explained?

K)

Problems with hand-based input/output

The current hand input devices suffer from the same delays that plague the head mount display systems, but the user's over-compensation is even more noticeable. Because there is typically some interaction with the hand and other objects, absolute position control is much more important here than it would be with head positioning, where relative motion is usually sufficient.

These devices are also extremely limited in their ability to generate any kind of tactile force or feedback to the user. Based on our research, even to perform gross manipulation tasks with a DataGlove is extremely difficult without some kind of sensory feedback. Any kind of fine manipulation is impossible. Though tactile feedback of some kind may be possible, the quality of this will very likely be extremely low and the cost extremely high for the foreseeable future. ^.

139

-4 Perhaps the major failing of the glove-based system is that it requires the

15 user to keep the hand and arm unsupported. This requires the user to employ both the agonist and antagonist muscle sets of the arm working against each other in order to perform any kind of complex task. The user actually is working harder at this than he would at pushing a real object because, in the case of a real object, at least one muscle group is at rest.

20 Further, because there is no true stable surface for the arm to rest against, any kind of control requires even more force between the muscle groups. Our experience demonstrated that a user of such a system when faced with any kind of gross manipulation tasks, could only be expected to use the system for five-minute periods with a large degree of exertion. Any

25 kind of extended activity was precluded.

As a consequence of these drawbacks, it is our expectation that the DataGlove and other similar interface devices will be replaced by more useful devices in the future. ■

► Vocabulary

agonist and antagonist muscle sets (1. 16) - two muscle groups which normally act in opposition to each other

Task 13

1 2 3

Read the text again and complete the table in note form. Problem of hand input device Consequences

Task 14

□ □ □

_□

1 2 3 4 5 6 7 8

The following pairs of words are taken from the text. In each case, say whether their meanings are similar (S) or opposite (0).

suffer from (1. 1) plague (1. 1) absolute (1. 4) true (1. 20) tactile (1. 8) sensory (1. 10) force (1. 8) exertion (1. 24) gross (I. 9) fine (1. 10) impossible (I. 11) precluded (1. 25) failing (1. 14) drawback (I. 26) wnrkine H. 1 81 at. rest (1.191

Language focus L

Classifying

The term 'classifying' means arranging objects in classes or groups according to shared characteristics. For example, the class of'animals' includes all living things that can feel and move about, such as fish and birds. Each of these subgroups is also a class in its own right, having shared characteristics.

Classifying, then, is a process of bringing order out of confusion by organizing information in a logical fashion. There are often several ways of classifying the same information.

1 From general to specific: focusing on the large or high-level category and

talking about its parts, that is from general to specific, the following expressions can be used:

is	is made up of
can be divided into	is composed of
is of	comprises
has	consists of

A general-to-specific classification will usually have singular main verbs, unless two or more things are being analysed simultaneously.

Examples:

1. The CPU is divided into three parts: the control unit, the arithmetic-logic unit, and memory.

2. The CPU has three parts: the control unit, the arithmetic-logic unit, and memory.

3. The CPU is made up of three parts: the control unit, the arithmetic-logic unit, and memory.

4. The CPU is composed of three parts: the control unit, the arithmetic-logic unit, and memory.

5. The CPU consists of three parts: the control unit, the arithmetic-logic unit, and memory.

2 From specific to general: what the smaller (or lower-level) components make when they are put together. This kind of classification uses the following expressions:

make up

form

constitute

maybe

can be

are classified as

142

Exercise 1

A specific-to-general classification will have plural verbs, because two or more lower-level categories are the focus of classification.

Examples:

1. [lie control unit, the arithmetic-logic unit, and memory are the three parts that make up the CPU.

2. The control unit, the arithmetic-logic unit, and memory arc the three parts that form the CPU.

Using the diagram below, complete the paragraph on the following page.

A Computer system

143

A computer, has four basic components: input, processor, memory, and

output. The CPU consists of two parts: the 1 , which directs

and controls the signals and commands inside the processor, and the

I unit, which does the arithmetic operations and the

decision-making operations. While the is made up of a

4 5 6 ]

., a

_, the'

, a , and a

. is composed of -

In a computer, internal memory or

locations inside the computer, whereas

- refers to the storage refers to the storage

embodied in the peripherals.

14

- may be divided into

16

_)and

17

,18

_).The.

. devices can be either a

22

21

-, a.

_,or a.

These devices enter information into the computer. After the processor has

operated on it, the'.

. devices display the results of the

25

, or store

24

or a

computations on either a

them on tape or disk for future use.

Exercise 2

Refer back to the text on C language (Unit 4, page 46) and complete the diagram.

144

Machine translation

Start-up

Task 1 Decide whether the following sentences are true (T) or false (F):

1. D Some machine translation (MT) systems produce good translations.

2. Q It is difficult to compare different MT systems.

3. Q The easiest way to evaluate any machine translation of a given text is to

compare it to a human translation of the same text.

Task 2

Reading

Read the text on the following page and check your answers to Task 1.

145

Lost in the machine translation

para para

1 It isn't just users who have trouble 6 with evaluation. Elliott Macklovitch, of Canada, described an evaluation of a large commercial MT system, in which he analysed the error performance of a series of software updates only to find - as the system's suspicious development manager had feared -that not only had there been no significant improvement, but the latest release was worse.

You can go out right now and buy a machine translation system for anything between £100 and £100,000. But how do you know if it's going to be any good? The big problem with MT systems is that they don't actually translate: they merely help translators to translate. Yes, if you get something like Metal (very expensive) or GTS (quite cheap) to work on your latest brochure, they will churn out something in French or whatever, but it will be pretty laughable stuff.

All machine-translated texts have to be extensively post-edited (and often pre-edited) by experienced translators. To offer a useful saving, the machine must make the time the translator spends significantly less than he or she would have taken by hand.

Inevitably, the MT manufacturers' glossies talk blithely of 'a 100 per cent increase in throughput', but scepticism remains. Potential users want to make their own evaluation, and that can tie up key members of the corporate language centre for months.

A few weeks ago, translators, system developers, academics, and others from Europe, the US, Canada, China, and Japan met for the first time in a Swiss hotel to mull over MT matters. A surprisingly large number of European governmental and corporate organizations are conducting expensive and elaborate evaluations of MT, but they may not produce 'buy or don't buy' results.

Take error analysis, a fancy name for counting the various types of errors the MT system produces. You might spend five months working out a suitable scoring scheme - is correct gender agreement more important than correct number?-and totting up figures for a suitably large sample of text, but what do those figures mean? If one system produces vastly more errors than another, it is curiously inferior. But suppose they produce different types of error in the same overall numbers: which type of error is worse? Some errors are bound to cost translators more effort to correct, but jt requires a lot more work to find out which.

And bugs are still common. Using a 7 'test suite' of sentences designed to see linguistic weaknesses, researches in Stuttgart found that although one large system could cope happily with various complex verb-translation problems in a relative clause, it fell apart when trying to do exactly the same thing ' in a main clause. Developers are looking for bigger, better test suites to hop to keep such bugs under control.

Good human translators produce good translations; all MT systems produce bad translations. But just what is c< good translation? One traditional assessment technique involves a bunch of people scoring translations on various scales for ^/intelligibility ('Does this translation into English make sense as a piece of English?'); accuracy ('Does this piece of English give the same information as the French original?'); style, and so on. However, such assessment is expensive, and designing the scales is something of a black art.

Properly designed and integrated MT systems really ought to enhance the translator's life, but few take this on trust. Of course, they do things differently in Japan. While Europeans are dabbling their toes and most Americans deal only in English, the Japanese have gone in at the deep end. The Tokyo area already sports two or three independent MT training schools where, as the eminent Professor Nagao casually noted in his presentation, activities are functioning with the efficiency of the Toyota production line. We're lucky they're only doing it in Japanese.

146

k3

Each of the sentences below (except one) summarizes an individual paragraph of the text. Order the sentences so that they form a summary of the text. One of the sentences contains information which is not in the text. Which one?

1 LJ The developers ofMT systems have also had problems evaluating their

systems.

2 LJ Many European organizations are evaluating MT, but the results may not

be conclusive.

3 LJ Assessing machine translations as good or bad is very difficult because

such judgements cannot be made scientifically.

4 ED It is time-consuming for potential users to test the MT manufacturers'

claims that their products double productivity.

5. LJ Better tests are needed to monitor linguistic weaknesses in MT systems.

6. ED All machine translations need to be edited by a human translator.

7. EH A reliable MT system is unlikely to be available this century.

8. LJ The price of MT systems varies greatly and none actually translates.

9. LJ The Japanese have a few independent MT training schools, which are

said to be very efficient.

10 LJ Analysing the errors made by MT systems is inconclusive because it may only show that different systems produce similar numbers of different error types.

sk4

1

2 3 4 5 6

Match each of the following verbs from the text with the expression that has a similar meaning:

a b c d e f

add up

think carefully about

manage successfully

produce large amounts of

fail

occupy the time of

churn out (para. 1) tie up (para. 3) mull over (para. 4) tot up (para. 5) cope with (para. 7) fall apart (para. 7)

sk5

1 2 3 4 5 6 7 8

Using the paragraph reference given, find words or phrases in the text which have a similar meaning to:

ridiculous (para. 1) colour brochures (para. 3) casually (para. 3) sure to (para. 5) group (para. 8) mysterious ability (para. 8) experimenting in a small way (para. 9) invested heavily (para. 9)

isk6

Speaking

Look at these sentences. Discuss why a machine might find them difficult to translate.

The sun set at 9 p.m.

We set off for London in the morning.

The VCR is on the television set.

I bought a set of six chairs.

He set a book on the table.

She had her hair set for the party.

Can you think of other examples where this kind of problem occurs?

147

AI and expert systems

Task 7

T ask 8

1 2

IS)

1 2 3 4 5 6 7 8

Listening

/

You are going to hear a conversation in which David, a graduate student doii research in the field of artificial intelligence, explains to a friend, Kevin, what AI and expert systems are. Before you listen, try to write short definitions to explain:

artificial intelligence expert systems

Now listen to the conversation and modify your definitions as necessary.

Listen again to the recording and answer these questions:

Does visual perception require intelligence when done by humans?

What two categories of task are mentioned in relation to AI programs?

Which category of task is AI more successful at?

What is the relationship between AI and expert systems?

What examples of existing expert systems are mentioned?

In what way do expert systems imitate human experts?

Why does the Japanese system have two parallel inference engines?

What is the function of inference trees?

Task 9 o Read this adapted extract from the tapescript and fill in the gaps with the

missing words.

used for?

._! s.

kevin: What are e._.._.

., Up to now they've been _, electronic fault finding,

david: They're built for commercial a.

used for a variety of tasks - medical d

machine translation, and so on. But the point about them is that you can i them about how they came to a particular c *.

148

kevin: So, in that respect, they imitate human experts.

david: Yes. I read recently about a Japanese system that can be used by

1 to draw conclusions about new legal cases. It refers to

d of statutory laws and legal precedents and is able to see

similarities in the r — processes used to decide each case - exactly

as a s lawyer would.

kevin: How can it do that?

david: The system has two reasoning mechanisms, known as i "

13

e , i

, which work in p

One operates on the written

laws, the other operates on the legal precedents. They draw all the possible conclusions and then output them in the form of i 1_⁴

15

Now listen again to the recording and check your answers.

Task 10

Reading

Read quickly through the text which follows and note down the answers to the following questions:

1. What does the expert system ROI do?

2. How did Scott French 'clone' Jacqueline Susann?

3. What other applications of AI are mentioned in the text?

One tough cookie

The software division of Mrs. Fields Cookies, Fields Software Group, has sold a version of its AI-based Retail Operations Intelligence 5 system to fast-food giant Burger King Corp. The expert system, called ROI, assists in the management of franchised or multiple-location retail operations

10 by creating work schedules, recommending marketing tactics, and assisting in personnel hiring. Fields has been successful with this package and has started

15 commercializing it. Now Burger King is developing its own expert system in an attempt to outperform its hamburger competitor McDonald's. Maybe it can clone

20 Ronald McDonald's expertise.

AI waxes poetic

Cloning a well-known figure is no joke. Just This Once is a new novel

25 30

35

40

45

making the rounds in the publishing world. It was written by Scott French, who claims that 10% of the novel was written by him, 25% was created by an AI program he created to imitate novelist Jacqueline Susann, and the remainder was a collaborative effort between himself and the computer. Susann, who died in 1974, wrote the definitive trash novel -and one of history's all-time bestsellers - Valley of the Dolls. French used Nexpert Object and took development lessons from Bechtel AI Institute to program his system with hundreds of formulas he had developed regarding Susann's essential plots and characterizations; it created a 350-page novel, which some in the literary community are

149

"^ calling 'computerized literary ghost- 60 writing'. While not all the reviews on his methodology are positive

50 (some think it is a violation of Susann's intellectual property), French claims, 'I don't think you 65

can copyright the way a person thinks.' If French gets a book deal,

55 this would be the first computer- generated novel ever published. Just this once, indeed. • 70

Hundreds of other places are employing AI; some of the

applications may seem mundane while others are intriguing. From expert systems that help you plan your garden to voice systems that help doctors treat critically injured patients in emergency rooms to natural language front-ends for multimedia systems that feature models, actors, and actresses, AI is being accepted in arenas outside those traditional realms of science and engineering. ■

Task 11

1 2 3 4 5 6 7 8

Answer these questions about the text.

What does the acronym ROI stand for?

Why is Burger King developing an expert system?

What kind of books did Susann write?

What percentage of his novel did French write jointly with his computer?

Has French's novel been well-received?

How does French justify his action?

Has French found a publisher for his book?

Where has AI traditionally been accepted?

Task 12 Choose the definition

1. franchised (1. 8)

2. hiring (1.12)

3. outperform (1. 17)

4. clone (1.19)

5. making the rounds (1.

6. trash (1. 34)

7. ghost-writing (1. 47)

8. mundane (1. 60)

that best expresses the meaning of the word or phrase.

a b c

licensed to sell another company's products

individual

specially selected

a managing

b employing

c training

a do better than

b remove from the top position

c survive longer than

a copy

b use to one's own advantage

c make people laugh at

23) a circulating

b making no progress

c making a bad impression

a printed on cheap paper

b popular

c of poor quality

a writing under someone else's name

b writing stories intended to frighten

c writing for someone who is dead

a ordinary

b simple

c world-wide

150

sk 13

Writing

Read this summary of the first paragraph of the text on page 149, then compare it to the original.

The software division of Mrs. Field's Cookies has sold a version of its Al-basecl system for assisting in the management of retail operations to Burger King. who are now developing their own system in an attempt to outperform McDonald's.

1. Note the information that has been omitted from the summary.

2. Look carefully at how the remaining information has been re-ordered and condensed.

3. Now complete the summary of the text, keeping it as concise as possible.

Word-play

isk 14

The clues below contain anagrams of words from this unit. Enter the words in the grid, then solve the anagram in the bold boxes to find the hidden word.

Expert systems have been used in medical.

. (ossadiing)

1

2. A set of instructions for making something, (lamurof)

3. Modernize, (dapetu)

4. A rival company, (torpetimoc)

5. Used of lines which are the same distance apart at any point, (alerlapl)

6. Assessment, (tualiaveno)

7. Having formal permission to sell another company's goods in a particular geographical area, (danrifcesh)

8. An exact copy, (ecnol)

9. Machines are still not very good at doing this, (nartgslanti)

Hidden word clue

This kind of engine is one of two reasoning mechanisms in an expert system.

151

152

Language focus M

_{Cause and effect}

Understanding the different ways of expressing the relationship between the causes and the effects of an action is very important when you are reading English. This cause-effect relationship is commonly used in texts about ■ computing.

Before we look at some of the ways of expressing cause and effect, note carefully this important distinction.

We can mention the cause before the effect.

Example:

(cause) (effect)

Dust often causes the recording condition of disks to deteriorate.

We can mention the effect before the cause.

Example:

(effect) (cause)

Deterioration in the recording condition of disks is often due to dust.

There are many different ways of expressing cause and effect. 1 Verbs linking cause and effect:

result	cause
produce	result in
allow	result from
prevent	bring about
enable

Examples:

1. The introduction of computer technology brought about significant change! in office routines, (cause -> effect)

2. Computers can create artificial objects in their memories. This allows developers to test product design without actually creating a real prototype. (cause -> effect)

3. The extensive use of computers in schools is resulting in a new generation c computer-literate students, (cause -► effect)

4. The problems were caused by the volume of network traffic* (effect ■*— cause)

Note: * See Language focus H for an explanation of the passive used in example 4.