# Notes from Monday, November 13 # A simple rule-based system for natural language processing import string #---------------------------------------------------------------------------------- class Movie: def __init__(self, title, year, director, actors): self.title = title self.year = year self.director = director self.actors = actors def __str__(self): return "%s (%d)" % (self.title, self.year) def info(self): print "%s (%d)" % (self.title, self.year) print "Directed by" , self.director print "Starring", string.join(self.actors, ", ") #---------------------------------------------------------------------------------- # cleanup removes punctuation characters from a string def cleanup(phrase): chars = [ch for ch in phrase if ch not in string.punctuation] return string.join(chars, "") #---------------------------------------------------------------------------------- class Rule: def __init__(self, pattern, action): assert pattern.count('...') <= 1, "no more than one '...' symbol allowed" self.pattern = pattern self.action = action def __str__(self): return "%s" % self.pattern def match(self, phrase): phrase = cleanup(phrase) wordsLeft = string.split(phrase) symbolsLeft = self.pattern substitutions = [] while len(symbolsLeft) > 0 and len(wordsLeft) > 0: firstSymbol = symbolsLeft[0] symbolsLeft = symbolsLeft[1:] if firstSymbol == '...': numMatched = len(wordsLeft) - len(symbolsLeft) wordsMatched = wordsLeft[:numMatched] substitutions.append(string.join(wordsMatched)) wordsLeft = wordsLeft[numMatched:] elif firstSymbol == '#': try: substitutions.append(eval(wordsLeft[0])) wordsLeft = wordsLeft[1:] except: return None elif type(firstSymbol) is list and wordsLeft[0] in firstSymbol: substitutions.append(wordsLeft[0]) wordsLeft = wordsLeft[1:] elif firstSymbol == '_': substitutions.append(wordsLeft[0]) wordsLeft = wordsLeft[1:] elif firstSymbol == wordsLeft[0]: wordsLeft = wordsLeft[1:] else: return None if len(symbolsLeft) > 0 or len(wordsLeft) > 0: return None else: return substitutions def apply(self, phrase): subs = self.match(phrase) if subs is not None: result = self.action(*subs) return result #---------------------------------------------------------------------------------- class MovieMeister: def __init__(self): self.database = [] self.rules = [] def addRule(self, pattern, action): rule = Rule(pattern, action) self.rules.append(rule) def addMovie(self, title, year, director, actors): movie = Movie(title, year, director, actors) self.database.append(movie) def process(self, phrase): for rule in self.rules: if rule.match(phrase): result = rule.apply(phrase) return result return None def interact(self): print "Welcome to the Movie Meister!" while True: input = raw_input("May I help you? ") if input in ['quit', 'bye', 'goodbye']: break results = self.process(input) if results is None: print "I don't understand." elif results == []: print "I don't know." else: for x in results: print x print "Come back again soon!" #---------------------------------------------------------------------------------- # The knowledge base mm = MovieMeister() mm.addMovie("Amarcord", 1974, "Federico Fellini", ["Nagali Noel", "Bruno Zanin", "Pupella Maggio", "Armando Drancia"]) mm.addMovie("The Godfather", 1972, "Francis Ford Coppola", ["Marlon Brando", "Al Pacino", "James Caan", "Robert Duvall", "Diane Keaton"]) mm.addMovie("The Big Easy", 1987, "Jim McBride", ["Dennis Quaid", "Ellen Barkin", "Ned Beatty", "Lisa Jane Persky", "John Goodman", "Charles Ludlam"]) mm.addMovie("Boyz in the Hood", 1991, "John Singleton", ["Cuba Gooding Jr.", "Ice Cube", "Larry Fishburne", "Tyra Ferrell", "Morris Chestnut"]) mm.addMovie("Psycho", 1960, "Alfred Hitchcock", ["Anthony Perkins", "Janet Leigh", "Vera Miles"]) mm.addMovie("Unforgiven", 1992, "Clint Eastwood", ["Clint Eastwood", "Gene Hackman", "Morgan Freeman", "Richard Harris"]) mm.addMovie("High Plains Drifter", 1973, "Clint Eastwood", ["Clint Eastwood", "Verna Bloom", "Marianna Hill"]) mm.addMovie("Dirty Harry", 1971, "Don Siegel", ["Clint Eastwood", "Harry Guardino", "Reni Santoni", "John Vernon"]) mm.addMovie("Annie Hall", 1977, "Woody Allen", ["Woody Allen", "Diane Keaton", "Tony Roberts"]) mm.addMovie("Star Wars", 1977, "George Lucas", ["Mark Hamill", "Harrison Ford", "Carrie Fisher", "Alec Guinness"]) #---------------------------------------------------------------------------------- # Pattern/action production rules pattern1 = ['What', 'movies', 'were', 'made', 'between', '#', 'and', '#'] def action1(year1, year2): return [movie.title for movie in mm.database if year1 <= movie.year <= year2] mm.addRule(pattern1, action1) #---------------------------------------------------------------------------------- pattern2 = ['What', 'movies', 'did', '...', ['act', 'star', 'appear'], 'in'] def action2(actor, verb): return [movie.title for movie in mm.database if actor in movie.actors] mm.addRule(pattern2, action2) #---------------------------------------------------------------------------------- pattern3 = ['What', 'movies', 'were', 'made', ['in', 'before', 'after', 'since'], '#'] def action3(prep, year): if prep == 'in': return [movie.title for movie in mm.database if movie.year == year] elif prep == 'before': return [movie.title for movie in mm.database if movie.year < year] elif prep == 'after' or prep == 'since': return [movie.title for movie in mm.database if movie.year > year] else: return None mm.addRule(pattern3, action3) #---------------------------------------------------------------------------------- pattern4 = ['Who', ['directed', 'wrote'], '...'] def action4(verb, title): if verb == 'wrote': return [] elif verb == 'directed': return [movie.director for movie in mm.database if movie.title == title] mm.addRule(pattern4, action4) #---------------------------------------------------------------------------------- pattern5 = ['Who', ['starred', 'acted', 'appeared'], 'in', '...'] def action5(verb, title): for movie in mm.database: if movie.title == title: return movie.actors return [] mm.addRule(pattern5, action5) #---------------------------------------------------------------------------------- pattern6 = ['When', 'was', '...', 'made'] def action6(title): return [movie.year for movie in mm.database if movie.title == title] mm.addRule(pattern6, action6) #---------------------------------------------------------------------------------- pattern7 = ['What', 'movies', 'did', '...', 'direct'] def action7(director): return [movie.title for movie in mm.database if movie.director == director] mm.addRule(pattern7, action7) #---------------------------------------------------------------------------------- # Example # # >>> mm.interact() # Welcome to the Movie Meister! # May I help you? What movies did Francis Ford Coppola direct? # The Godfather # May I help you? What movies did Clint Eastwood direct? # Unforgiven # High Plains Drifter # May I help you? What movies did Clint Eastwood star in? # Unforgiven # High Plains Drifter # Dirty Harry # May I help you? Who directed Dirty Harry? # Don Siegel # May I help you? What year was Dirty Harry made? # I don't understand. # May I help you? When was Dirty Harry made? # 1971 # May I help you? What movies were made between 1970 and 1980? # Amarcord # The Godfather # High Plains Drifter # Dirty Harry # Annie Hall # Star Wars # May I help you? Who acted in The Godfather? # Marlon Brando # Al Pacino # James Caan # Robert Duvall # Diane Keaton # May I help you? What movies did Diane Keaton appear in? # The Godfather # Annie Hall # May I help you? Who directed Annie Hall? # Woody Allen # May I help you? Who wrote Annie Hall? # I don't know. # May I help you? What movies were made after 1980? # The Big Easy # Boyz in the Hood # Unforgiven # May I help you? goodbye # Come back again soon!