Transcript Lara`s slides ()

The Construction of a Pun
Generator for Language Skills
Development
Humor Generation
SoSe 2010
Lourdes Lara Tapia
Overview








Introduction.
Early pun generators.
JAPE.
STANDUP.
STANDUP in the Praxis.
Evaluation
Conclusion.
References.
June 29, 2010
Humor Generation
2
Introduction
 What is a Pun Generator?
 A pun Generator is a Computer Program which
generates jokes.
 What is a Joke?
 It is a short text which provoke laughter.
 A joke has normally a Punchline.
 There are different kind of Jokes:
 Punning riddles
June 29, 2010
Humor Generation
3
Introduction
 A punning riddle is a simple question-answer
joke in which the answer makes a play on
words:
 What do you call a good bye that has a tooth?
 A saw long.
June 29, 2010
Humor Generation
4
Introduction
 What kind of ambiguity is used here?
Phonetic similarity
Semantic relation
 What do you call a good bye that has a tooth?
Synonym
 A saw long.
Meronym
Homophone
 A So long
June 29, 2010
Humor Generation
5
Early pun generators
 Raskin (1985):
 Incongruity Theory.
 Lesard & Levison (1992):
 VINCI: Tom Swift
 “we must hurry”, said Tom Swiftly.
 “I hate Math”, Tom added
 Binsted & Ritchie (1994):
 JAPE:
 Punning riddle uses phonological and semantical ambiguity
 Used a large lexicon (WordNet)
 Properly controlled evaluation of the output was carried out.
June 29, 2010
Humor Generation
6
Early pun generators
 Venour (1999):
 The Homonym Common Phrase Pun (HCPP).
 A one-sentence set-up &
 A punning punchline.
 Mechanismus are similar to those used in JAPE
 McKay (2002):
 WISCRAIC:
 Simple puns in 3-different linguistic forms:
 Question-answer, single and two-sentences sequence.
 Support 2nd-language learning
June 29, 2010
Humor Generation
7
Early pun generators
 Nijholt (2003):
 Communication with machines.
 Stock et al. (2005):
 HAHAcronym:
 Acronym  funny concepts
 Concept  funny Acronym
 Mihalcea & Strapparava (2006):
 Techniques to humor recognition:
 Humurous and non-humorous.
June 29, 2010
Humor Generation
8
JAPE
 Joke Analysis and Production Engine.
 What is JAPE?
 Computer Program
 In Prolog by Binsted in 1994.
 Several Version
 JAPE-1 (pilot version) & JAPE2
 JAPE-3 & JAPE-4 (more flexible dictionary module)
  STANDUP in 2008.
June 29, 2010
Humor Generation
9
JAPE
 JAPE produced short texts
 punning
riddles:
 What is the difference between a pretty glove and a
silent cat?
 One is a cute mitten, the other is a mute kitten.
 The Jokes were reliably distinguished from Non-
Jokes.
 The best of these were published in joke books
for children.
June 29, 2010
Humor Generation
10
JAPE
 The three main strategies used to create
phonological ambiguity:
 syllable substitution,
 word substitution &
 Metathesis
 Joke-construction mechanisms.
 Very similar to those in STANDUP
June 29, 2010
Humor Generation
11
JAPE
 Deficiencies:
 Few parameters available for variation.
 There was no way to guide the software.
 No real user interface.
 The search for suitable words could be slow,
unintelligent and exhaustive.
 Good intelligible jokes was very small.
 No facilities to compare words for phonological or
semantically ambiguity.
June 29, 2010
Humor Generation
13
STANDUP
 System To Augment Non-speakers Dialogue Using
Puns.
 This Program is aimed at young children, and lets them




play with words and phrases by building punning riddles
through a simple interactive user-interface.
Allow young children to explore the language.
Children with Complex Communication Needs (CNN).
Punning riddle.
“Language playground”
June 29, 2010
Humor Generation
14
STANDUP
Schema
Description Rules
Templates
Phrasal
Question or
Answer
Header
Lexical Precondition
Header
Question Spec.
Answer Spec.
Preconditions
Header
Template Spec.
Body
Keywords
June 29, 2010
Humor Generation
15
Fig. http://www2.hawaii.edu/~bergen/papers/humor-IEEE.pdf
June 29, 2010
Humor Generation
16
STANDUP
 What do you call a shout with a window?
 A computer scream.
June 29, 2010
Humor Generation
17
STANDUP
11 Schema
(kind of joke)
Description Rules
Templates
Lexical Precondition:
Header:
Newelan2(NP,A,B,HomB)
Nouncomp(NP,A,B),
Homoph(B,HomB),
Noun(HomB)
Question Spec.:
Answer Spec.:
{Shareprop (NP,HomB)}
{phrase (A,HomB)}
Header:
Shareprop {X,Y}
Preconditions:
Meronym(X,MerX),
Syn(Y,SynY)
Keywords:
Template Spec.:
[NP,HomB]
[merHyp, MerX, SynY]
June 29, 2010
Phrasal (finish touches)
Question (What is the diff…?)
Answer (They’re both…)
Humor Generation
Header
Body
18
STANDUP
11 Schema
(kind of joke)
Description Rules
Templates
Header:
Lexical Precondition:
Header:
Newelan2(NP:computer screen,
A: computer,
B: screen,
HomB: scream)
Nouncomp(NP,A,B),
Homoph(B,HomB),
Noun(HomB)
Shareprop
{computer screen,
scream}
Question Spec.:
{Shareprop (computer screen,
scream)}
Preconditions:
Answer Spec.:
{phrase (computer, scream)}
Meronym(computer screen, window),
Syn(scream, shout)
Keywords:
Template Spec.:
[NP,HomB]
[merSyn, window, shout]
June 29, 2010
Phrasal (finish touches)
Question
Question (What
is the diff…?)
What do you call a shout with a window?
Answer (They’re both…)
Humor Generation
Header
A shout with
a window
[merSyn,
window,
shout]
Body
Body
What do you call a
NP(shout) with a
NP(X,Y)
NP(window)
19
STANDUP-Lexicon
 WordNet as JAPE +
 Phonetic similarity.
 Speech Output.
 Picture Support.
 Topics.
 Familiarity of words.
 Vocabulary restriction.
June 29, 2010
Humor Generation
21
STANDUP-Facilities
 Joke telling:
 VOCA: Voice-Output Communication Aid.
 assists people who are unable to use natural speech to express
their needs and exchange information with other people during a
conversation.
 User Profiles:
 Username.
 Two kind of data:
 Option settings.
 Personal Data.
 Standard Package:
 Beginner
 Touchscreen-user.
June 29, 2010
Humor Generation
22
STANDUP-Facilities
 Logging:
 Logged in a Disc file:
 Allows researchers to study usage as required.
 Log player
 Dump the simulated re-runs into a video file.
June 29, 2010
Humor Generation
23
STANDUP-Software
 ..\STANDUP Simple.bat
June 29, 2010
Humor Generation
24
STANDUP-Evaluation
 Evaluate the effectiveness of the software.
 No ambitious but qualitative study.
 A group of 9 children (8-12years old) with cerebral palsy.
 Scholars used the software spontaneously,
 Found the “Tell the jokes-function” amazing and
 Re-told the jokes afterwards.
 8 children reacted very positively
 1 of the older boys complained about the quality of the Jokes.
 Anecdotal evidence: Children’s communication had
improved.
June 29, 2010
Humor Generation
25
STANDUP-Evaluation
 In the post-testing:
 The Preschool and Primary Inventory of
Phonological Awareness, PIPA, showed no sign
of improved.
 Clinical Evaluation of Language Fundamentals,
CELF, only the older boy, who complained,
showed no sign of improved.
June 29, 2010
Humor Generation
26
Conclusion
 Humor is one of the most sophisticated forms
of human intelligence.
 On the cognitive side humor has two very
important properties:
 it helps getting and keeping people’s attention.
 it helps remembering.
 On the artificial intelligence side
computational humor is a challenge with
implications for many classical fields.
June 29, 2010
Humor Generation
27
Conclusion
 The development of all its facets is not
something for the near future, the
phenomena are too complex.
 Simple puns, at least, can be modelled
formally, and can be generated by a program.
 The software is definitely usable for a
practical application by children with
communication disabilities to develop their
linguistic skills.
June 29, 2010
Humor Generation
28
Discussion
 Questions

 Opinion or

 Comments
June 29, 2010
Humor Generation
29
Thank you for your attention
June 29, 2010
Humor Generation
30
References









Binsted, K. 1996. Machine humour: An implemented model of puns. Ph. D. thesis,
University of Edinburgh, Edinburgh, Scotland.
Binsted, K., H. Pain, and G. Ritchie. 1997. Children's evaluation of computer generated
punning riddles. Pragmatics and Cognition 5 (2), 305-354.
Manurung, R., G. Ritchie, H. Pain, A. Waller, D. O'Mara, R. Black (2008). The construction
of a pun generator for language skills development. Applied Artificial Intelligence, 22(9) pp.
841-869.
Ritchie, G. 2001. Current directions in computational humour. Artificial Intelligence Review
16 (2), 119-135.
Ritchie, G. 2003. The JAPE riddle generator: technical specification. Informatics Research
Report EDI-INF-RR-0158, School of Informatics, University of Edinburgh, Edinburgh.
Stock, O. and C. Strapparava. 2003. HAHAcronym: Humorous agents for humorous
acronyms. Humor: International Journal of Humor Research 16 (3), 297-314.
http://www.csd.abdn.ac.uk/~gritchie/
http://www.csd.abdn.ac.uk/research/standup/software.php
http://www.csd.abdn.ac.uk/research/standup/downloads/UserManual.html
June 29, 2010
Humor Generation
31