1. LING 388 Language and Computers Lecture 2 9/04/03 Sandiway FONG

2. Administrivia Course Webpage http://www.u.arizona.edu/~sandiway/ling388/ Course Webpage http://www.u.arizona.edu/~sandiway/ling388/ If you don’t have PC access… If you don’t have PC access…  Computer Laboratory  SBS 224  Mon-Fri 8am-6pm My email: My email:  sandiway@email.arizona.edu TA: TA:  Charles Lin (clin@email.arizona.edu) clin@email.arizona.edu

3. Administrivia

4. Next Tuesday (9th) Next Tuesday (9th)  A computer lab class  Change of location: SBS 224 First Homework First Homework  Out next Tuesday  Due the following Tuesday  Homeworks should be submitted electronically to the TA

5. Today’s Lecture Introduction to PROLOG Introduction to PROLOG  PROLOG = Programming in Logic  Based on a subset of first order logic  Advantages: Looks like logic not C/C++/JavaLooks like logic not C/C++/Java Very high-level and easy to useVery high-level and easy to use –no nasty primitive programming stuff like pointers or memory allocation or variable typing

6. Today’s Lecture We’ll find it useful for: We’ll find it useful for:  Formalizing or defining concepts  Also happens to be a convenient high-level programming language Background Background  Resolution theorem proving (Robinson, 65)  Invented by Colmerauer and Kowalski in the early 70s  Designed to support natural language processing… grammar rules

7. Highlights Backtrack mechanism for non-determinism Backtrack mechanism for non-determinism  Can return more than one answer Depth-first search Depth-first search  Memory efficient way to look for answers Unification Unification  Mechanism used to assign values to variables Negation-as-failure (to prove) Negation-as-failure (to prove)  Limited form of logical negation Definite Clause Grammar (DCG) notation Definite Clause Grammar (DCG) notation  Useful for writing grammar rules

8. Task 1 Install Prolog on your personal computer Install Prolog on your personal computer Free download Free download  SWI-Prolog (University of Amsterdam)  Platforms supported:  Windows, Unix/Linux, MacOS X,…  Website: Website:  http://www.SWI-Prolog.org/ http://www.SWI-Prolog.org/ Will be also available on the machines in Computer Lab SBS 224 Will be also available on the machines in Computer Lab SBS 224

9. Prolog: Basic Data Structures Constants Constants  Names (begin with lower case letter)  Examples: john mary theMan x  Exception: May begin with an upper case letter if enclosed in single quotesMay begin with an upper case letter if enclosed in single quotes –Example: ‘John’ ‘X’  Numbers  Examples: 100 3.14159

10. Prolog: Basic Data Structures Variables (begin with an upper case letter) Variables (begin with an upper case letter)  Examples: X Y23 Variable John  Exception:  May also begin with an underscore  Examples: _ _12 _X _ is special: called the “anonymous variable”_ is special: called the “anonymous variable”  Initially has no value (unbound)  May only be assigned a value once  X = john means “X is bound to the constant john”means “X is bound to the constant john”  cf. X := 1; Y := 2; X := 2 … in many languages

11. Prolog: Basic Data Structures Variables Variables  Scope of variable is a clause meaning: if some variable is mentioned more than once in a clause, it’s the same variablemeaning: if some variable is mentioned more than once in a clause, it’s the same variable clause = smallest independent program fragmentclause = smallest independent program fragment  Example: animate(X) :- man(X). read as: “X is animate if X is a man”read as: “X is animate if X is a man” both Xs denote the same variableboth Xs denote the same variable  Example: animate(X) :- woman(X). the X here is independent of the X in the previous clausethe X here is independent of the X in the previous clause

12. Prolog: More Data Structures Lists (delimited by square brackets) Lists (delimited by square brackets)  Examples:  [] empty list  [1] list with one element ‘1’  [1,2,3] list with three elements 1,2 and 3  Alternative notation (using | ):  [1|[2,3]] same as [1,2,3]  [1|[]] same as [1]  Other Examples:  [X,mary] list containing a variable X and a constant mary  [1|X] list with 1st element 1 and remainder denoted by a variable  [[1,2],[3,4],[5,6]] list with 3 elements, each a list

13. Prolog: More Data Structures Structures (functor/argument(s)) Structures (functor/argument(s))  functor must be a name  argument can be anything… name, variable or another structure  Examples:  np(john) structure with functor np and a constant argument john structure with functor np and a constant argument john  s(np,vp) structure with functor s and 2 arguments np and vpstructure with functor s and 2 arguments np and vp  s(np(john),vp(X)) - nested structures- nested structures Note: We can represent parse trees using nested structures Note: We can represent parse trees using nested structures

14. Prolog: Clauses Clauses may be program or database facts Clauses may be program or database facts  facts resemble structures + ‘.’ (period)  Examples:  man(john). woman(mary).  modal(should). modal(could). aux(be). aux(do).  lessThanTwo([]). lessThanTwo([_]).

15. Prolog: Clauses Clauses may be rules with conditions Clauses may be rules with conditions  General form: A :- B, C  In a rule:  the connective :- means “if”  the connective, means “and”  A is the head of the clause  B, C is the body of the clause  Reading: “A is true if B is true and C is true”

16. Prolog: Clauses Examples: Examples:  animate(X) :- man(X).  “X is animate if X is a man”  animate(X) :- woman(X).  “X is animate if X is a woman”  Note: two clauses for predicate animate/1 Notation: /1 means arity 1 - “has one argument”Notation: /1 means arity 1 - “has one argument”  Implies there are two ways to satisfy animate/1  Two clauses form a disjunctive definition for animate/1

17. Prolog: Clauses Alternative notation: Alternative notation:  animate(X) :- man(X) ; woman(X).  connective ; means “or”  means “X is animate if X is a man or X is a woman”  Note the scope of X Negation ( \+ restricted to the body of a clause) Negation ( \+ restricted to the body of a clause)  Example:  Define a predicate contractedNegative/1 to be read as “(has a) contracted negative” contractedNegative(X) :- modal(X).contractedNegative(X) :- modal(X). contractedNegative(X) :- aux(X).contractedNegative(X) :- aux(X).  Exception: *mayn’t contractedNegative(X) :- modal(X), \+ X = may.contractedNegative(X) :- modal(X), \+ X = may.

18. Prolog: Unification Unification ( = ) Unification ( = )  mechanism for matching and valuing variables  Examples:  john = john “john unifies with john” Note: john = mary will failNote: john = mary will fail  X = john or john = X “john unifies with X” “X is bound to john” “john unifies with X” “X is bound to john” Note: unification is symmetricNote: unification is symmetric  noun(X) = noun(john) “unifies with X bound to john” Note: match the functors, then match the argumentsNote: match the functors, then match the arguments

19. Prolog: Queries Query ( ?- ) Query ( ?- )  query = fact prefixed by ?-  matches against facts and rules in the program database  Examples:  modal(should). modal(could).  ?- modal(X). X = should ;X = should ; X = couldX = could  Note: type ; to the Prolog interpreter to request more answers  Note: answers are returned according to program database order

20. Prolog Queries Queries without variable binding Queries without variable binding  for verification only - yes/no answer  Examples:  ?- modal(should).  Yes (Prolog interpreter response)  ?- modal(be).  No (Prolog interpreter response)

21. Prolog Queries Negative queries Negative queries  Examples:  ?- \+ modal(be).  Yes  ?- \+ modal(should).  No  Question:  What does the query ?- \+ modal(X). return?  Explain why