深入学习Prolog编程

需积分: 0 123 浏览量更新于2024-08-02 1 收藏 1.93MB PDF 举报

"《深入理解Prolog》是一本专注于实践Prolog编程的书籍，不只停留在理论层面。书中提供了几个可以直接运行的专家系统外壳，以及用于排序、搜索、自然语言处理和数值方程求解的程序。作者是Michael A. Covington、Donald Nute和Andr´e Vellino，他们来自美国乔治亚大学的人工智能项目。本书包含1.1到1.8章节，详细介绍了Prolog语言的基础和核心概念，如Prolog的哲学、工作原理、不同变体、知识库构建、统一与变量实例化、回溯机制、语法以及关系定义等。" 深入理解Prolog这一主题，首先需要掌握Prolog的基本思想。Prolog是一种基于逻辑推理的编程语言，它的核心理念是通过一系列的事实和规则，利用逻辑推理来解决问题。1.1章的"介绍Prolog"部分将引导读者理解Prolog的基本概念，阐述其与其他编程语言的区别。 1.2章"Prolog的工作原理"中，会详细介绍Prolog如何通过查询数据库式的事实和规则，使用反向链式推理来寻找答案。这种工作方式使得Prolog在处理复杂逻辑问题时特别有效。 1.3章"Prolog的变体"讨论了不同的Prolog实现和版本，每个可能具有独特的特性和优化，这对于选择合适的Prolog环境进行开发至关重要。 1.4章"实用知识库"部分，作者可能会讲述如何构建和组织Prolog中的知识库，以便于存储和检索信息，这对于构建专家系统来说是基础。 1.5章"统一与变量实例化"是Prolog的核心机制之一，它描述了如何通过变量匹配和实例化来确定事实或规则之间的关系。 1.6章"回溯"解释了当一条路径的推理失败时，Prolog如何回溯到之前的决策点，尝试其他可能的解决方案，这是Prolog解决非确定性问题的关键。 1.7章"Prolog语法"则会详细介绍Prolog程序的结构和语法规则，包括原子、变量、列表、结构等元素的表示和使用。最后，1.8章"定义关系"将教授如何在Prolog中定义和使用关系，这包括编写规则、使用剪枝避免冗余计算等。通过这些章节的学习，读者将能够深入理解Prolog的精髓，具备使用Prolog开发实际应用的能力，包括但不限于专家系统、自然语言处理和数值计算等领域。这本书对于想要提升Prolog编程技能的开发者来说，是一份宝贵的资源。

Authors’ manuscript 693 ppid September 9, 1995 Prolog Programming in Depth

Sec. 1.2. How Prolog Works

[

]

To prove that

is in the United States, prove that

is in Georgia.

[

]

To prove that Atlanta is in Georgia, do nothing.

We ask our question by issuing the instruction:

Prove that Atlanta is in the United States.

This calls procedure[1

], whichin turn calls procedure [2

], whichreturnsthe answer

“yes.”

Prolog has its own notation for representing knowledge. Our sample knowl-

edge base can be represented in Prolog as follows:

in_united_states(X) :- in_georgia(X).

in_georgia(atlanta).

Here

in_georgia

and

in_united_states

are PREDICATES — that is, they say things

about individuals. A predicate can take any ﬁxed number of ARGUMENTS (parame-

ters); for example,

female(sharon).

might mean “Sharon is female,” and

mother(melody,sharon).

might mean “Melody is the mother of Sharon.” A predicate that takes

arguments

(for any number

) is called an

–PLACE PREDICATE; thus we say that

in_georgia

in_united_states

, and

female

are ONE–PLACE PREDICATES, while

mother

is a TWO–

PLACE PREDICATE. A one–place predicate describes a PROPERTY of one individual; a

two-place predicate describes a RELATION between two individuals.

The number of arguments that a predicate takes is called its ARITY (from terms

like unary, binary, ternary, and the like). Two distinct predicates can have the same

name if they have different arities; thus you might have both

mother(melody)

meaning Melody is a mother, and

mother(melody,sharon)

, meaning Melody is the

mother of Sharon. We will avoid this practice because it can lead to confusion.

In some contexts a predicate is identiﬁed by giving its name, a slash, and

its arity; thus we can refer to the two predicates just mentioned as

mother/1

and

mother/2

Exercise 1.2.1

Give an example, in Prolog, of: a fact; a rule; a clause; a one–place predicate; a predicate

of arity 2.

Exercise 1.2.2

In the example above, we represented “in Georgia” as a property of Atlanta. Write a

Prolog fact that represents “in” as a relation between Atlanta and Georgia.

Exercise 1.2.3

How would you represent, in Prolog, the fact “Atlanta is at latitude 34 north and

longitude 84 west”? (Hint: More than one approach is possible. Second hint: It’s OK to

use numbers as constants in Prolog.)

Authors’ manuscript 693 ppid September 9, 1995 Prolog Programming in Depth

Introducing Prolog Chap. 1

1.3. VARIETIES OF PROLOG

An important goal of this book is to teach you how to write portable Prolog code.

Accordingly, we will stick to features of the language that are the same in practically

all implementations. The programs in this book were developed in Arity Prolog

and ALS Prolog on IBM PCs and Quintus Prolog on Sun workstations. Most of

them have also been tested in SWI–Prolog, LPA Prolog, Cogent Prolog, and Expert

Systems Limited’s Public Domain Prolog–2.

For many years, the defacto standardfor Prologwas the language describedby

Clocksin and Mellish in their popular textbook, Programming in Prolog (1981, second

edition 1984). This is essentially the language implemented on the DEC-10 by D.

H. D. Warren and his colleagues in the late 1970s, and is often called “Edinburgh

Prolog” or “DEC–10 Prolog.” Most commercial implementations of Prolog aim to be

compatible with it.

In 1995 the International Organization for Standardization (ISO) published an

international standard for the Prolog language (Scowen 1995). ISO Prolog is very

similar to Edinburgh Prolog but extends it in some ways. Our aim in this book is to

be as compatible withthe ISO standard as possible, but without using featuresof ISO

Prolog that are not yet widely implemented. See Appendix A for more information

about ISO Prolog.

Finally, we must warn you that this book is not about Turbo Prolog (PDC

Prolog), nor about Colmerauer’s Prolog II and Prolog III. Turbo Prolog is Prolog

with data type declarations added. As a result, programs run faster, but are largely

unable to examine and modify themselves. Colmerauer’s Prolog II and III are

CONSTRAINT LOGIC PROGRAMMING languages, which means they let you put limits

on the value of a variable before actually giving it a value; this makes many new

techniques available. The conceptsin this bookarecertainly relevant to Turbo (PDC)

Prolog and Prolog II and III, but the details of the languages are different.

Exercise 1.3.1

If you have not done so already, familiarize yourself with the manuals for the version

of Prolog that you will be using.

Exercise 1.3.2

In the Prolog that you are using, does the query ‘

?- help.

’ do anything useful? Try it

and see.

1.4. A PRACTICAL KNOWLEDGE BASE

Figure 1.1 shows a Prolog knowledge base that describes the locations of certain

North American cities. It deﬁnes a single relation, called

located_in

, which relates

a city to a larger geographical unit. The knowledge base consists of facts such as

Users of ESL Public Domain Prolog–2 must select Edinburgh–compatible syntax by adding the

line ‘

:- state(token class, ,dec10).

’ at the beginning of every program. Note that ESL Prolog–2 has

nothing to do with Colmerauer’s Prolog II.

Authors’ manuscript 693 ppid September 9, 1995 Prolog Programming in Depth

Sec. 1.4. A Practical Knowledge Base

% File GEO.PL

% Sample geographical knowledge base

/* Clause 1 */ located_in(atlanta,georgi a).

/* Clause 2 */ located_in(houston,texas) .

/* Clause 3 */ located_in(austin,texas).

/* Clause 4 */ located_in(toronto,ontari o).

/* Clause 5 */ located_in(X,usa) :- located_in(X,georgia).

/* Clause 6 */ located_in(X,usa) :- located_in(X,texas).

/* Clause 7 */ located_in(X,canada) :- located_in(X,ontario) .

/* Clause 8 */ located_in(X,north_americ a) :- located_in(X,usa).

/* Clause 9 */ located_in(X,north_americ a) :- located_in(X,canada).

Figure 1.1

A simple Prolog knowledge base.

“Atlanta is located in Georgia,” “Houston is located in Texas,” and the like, plus

rules such as “

is located in the United States if

is located in Georgia.”

Notice that names of individuals, as well as the predicate

located_in

, always

begin with lower–case letters. Names that begin withcapital letters are variables and

can be given any value needed to carry out the computation. This knowledge base

contains only one variable, called

. Any name can contain the underscore character

(

Notice also that there are two ways to delimit comments. Anything bracketed

and

is a comment; so is anything between

and the end ofthe line, like this:

/* This is a comment */

% So is this

Commentsareignoredbythecomputer;weusethemto addexplanatoryinformation

and (in this case) to number the clauses so we can talk about them conveniently.

It’s not clear whether to call this knowledge base a program; it contains nothing

that will actually cause computation to start. Instead, the user loads the knowledge

base into the computer and then starts computation by typing a QUERY, which is a

question that you want the computer to answer. A query is also called a GOAL.It

looks like a Prolog clause except that it is preceded by ‘

’ — although in most cases

the Prolog implementation supplies the ‘

’ and you need only type the goal itself.

Unfortunately, wecannot tell you how to use Prologonyourcomputer, because

there is considerable variation from one implementation to another. In general,

though, the procedure is as follows. First use a text editor to create a ﬁle of clauses

such as GEO.PL in Figure 1. Then get into the Prolog interpreter and type the special

query:

?- consult('geo.pl').

(Remember the period at the end — if you don’t type it, Prolog will assume your

query continues onto the next line.) Prolog replies

yes

Authors’ manuscript 693 ppid September 9, 1995 Prolog Programming in Depth

Introducing Prolog Chap. 1

to indicate that it succeeded in loading the knowledge base.

Two important notes: First, if you want to load the same program again

after escaping to an editor, use

reconsult

instead of

consult

. That way you

won’t get two copies of it in memory at the same time. Second, if you’re using

a PC, note that backslashes (

) in the ﬁle name may have to be written twice (e.g.,

consult('c:\\myprog.pl')

to load C:

MYPROG.PL). This is required in the ISO

standard but not in most of the MS–DOS Prologs that we have worked with.

As soon as

consult

has done its work, you can type your queries. Eventually,

you’ll be through using Prolog, and you can exit from the Prolog system by typing

the special query

?- halt.

Most queries, however, retrieve information from the knowledge base. You can type

?- located_in(atlanta,georgia).

to ask whether Atlanta is in Georgia. Of course it is; this query matches Clause 1

exactly, so Prolog again replies “yes.” Similarly, the query

?- located_in(atlanta,usa).

can be answered (or, in Prolog jargon, SOLVED or SATISFIED) by calling Clause 5 and

then Clause 1, so it, too, gets a “yes.” On the other hand, the query

?- located_in(atlanta,texas).

gets a “no” because the knowledge base contains no information from which the

existence of an Atlanta in Texas can be deduced.

We say that a query SUCCEEDS if it gets a “yes” answer, or FAILS if it gets a “no”

answer.

Besides answering yes or no to speciﬁc queries, Prolog can ﬁll in the blanks in

a query that contains variables. For example, the query

?- located_in(X,texas).

means “Give me a value of X such that

in(X,texas)

succeeds.”

And here we run into another unique feature of Prolog — a single query can

have multiple solutions. Both Houston and Austin are in Texas. What happens in

this case is that Prolog ﬁnds one solution and then asks you whether to look for

another. This continues until all alternatives are found or you stop asking for them.

In some Prologs, the process looks like this:

?- located_in(X,texas).

X = houston

More (y/n)? y

X = austin

More (y/n)? y

Authors’ manuscript 693 ppid September 9, 1995 Prolog Programming in Depth

Sec. 1.4. A Practical Knowledge Base

The “no” at the end means there are no more solutions.

In Arity Prolog, the notation is more concise. After each solution, the computer

displays an arrow (

). You respond by typing a semicolon (meaning look for more

alternatives) or by hitting Return (meaning quit), like this:

?- located_in(X,texas).

X = houston -> ;

X = austin -> ;

In Quintus Prolog and many others, there isn’t even an arrow; the computer just

pauses and waits for you to type a semicolon and then hit Return, or else hit Return

by itself:

?- located_in(X,texas).

X = houston ;

X = austin ;

Also,you’llﬁndit hardtopredictwhetherthecomputerpausesafter thelastsolution;

it depends partly on the way the user interface is written, and partly on exactly what

you have queried. From here on, we will present interactions like these by printing

only the solutions themselves and leaving out whatever the user had to type to get

the alternatives.

Sometimes your Prolog system may not let you ask for alternatives (by typing

semicolons, or whatever) even though alternative solutions do exist. There are two

possible reasons. First, ifyour query hasperformed any output ofits own, the Prolog

system will assume that you’ve already printed outwhateveryou wanted tosee, and

thus that you’re not going to want to search for alternatives interactively. So, for

example, the query

?- located_in(X,texas), write(X).

displays only one answer even though, logically, there are alternatives. Second, if

your query contains no variables, Prolog will only print “yes” once no matter how

many ways of satisfying the query there actually are.

Regardless of how your Prolog system acts, here’s a sure–ﬁre way to get a list

of all the cities in Texas that the knowledge base knows about:

?- located_in(X,texas), write(X), nl, fail.

The special predicate

write

causes each value of

to be written out;

starts a new

line after each value is written; and

fail

forces the computer to backtrack to ﬁnd all

solutions. We will explain how this works in Chapter 2. For now, take it on faith.

We say that the predicate

located_in

is NONDETERMINISTIC because the same

question can yield more than one answer. The term “nondeterministic” does not

mean that computers are unpredictable or that they have free will, but only that they

can produce more than one solution to a single problem.

Another important characteristic of Prolog is that any of the arguments of

a predicate can be queried. Prolog can either compute the state from the city or

compute the city from the state. Thus, the query

剩余515页未读，继续阅读

zxrgreat

粉丝: 1
资源: 4

深入学习Prolog编程

Prolog 介绍

深度的Prolog编程（Michael A. Covington）Prolog Programming in Depth (Michael A. Covington)

数据手册-SN75176B-datasheet.zip

基于javaswing酒店点餐系统

WinPE-26241.5000-ReFS-v3.14.wim

Kubernetes系统精讲 Go语言实战K8S集群可视化--第8章 【核心知识+原理分析】将应用和配置分离.zip

节能减排商品类别.doc

三废污染+区域经济数据整理-最新出炉.zip

评标委员会成员评标意见表.doc

E056AER部分文献及其复现数据与程序_共37篇-最新出炉.zip

最新资源

Kubernetes系统精讲 Go语言实战K8S集群可视化--第8章【核心知识+原理分析】将应用和配置分离.zip