ungodly early hour

1
Ontology of Humor, Emotion, and Persuasion

• Victor Raskin
• Purdue University, USA
• vraskin_at_purdue.edu

2
Plan of Presentation

• A Snapshot of Ontological Semantics
• Computational Humor
• Ontology of Emotion (by Katrina
• Triezenberg)
• Ontology of Persuasion
• Humor in Persuasion

3
Plan of Presentation

• A Snapshot of Ontological Semantics
• Computational Humor
• Ontology of Emotion (by Katrina
• Triezenberg)
• Ontology of Persuasion
• Humor in Persuasion

4
Ontological Semantics

• Ontology hierarchy of conceptual nodes
• Lexicon entries explained in terms of nodes
• Necessary modules Analyzer, Generator
• Basis for analysis into Text Meaning
  Representation (TMR)

5
The ontological concept INFORM
inform definition the event of asserting
something to provide information to
another person or set of persons is-a
assertive-act agent human theme
event instrument communication-device benefici
ary human
6
The Lexicon Entry for Say
The lexicon entry for say, related to inform, is
represented in simplified form as say-v1 syn-str
uc 1 root say as in Spencer cat
v said a word subj root var1 cat
n obj root var2 cat n
7
The Lexicon Entry for Say 2
The lexicon entry for say, related to inform, is
represented in simplified form as say-v1 syn-str
uc 2 root say as in Spencer cat
v said that it subj root
var1 rained cat n comp root var2
8
The Lexicon Entry for Say 3
The lexicon entry for say, related to inform, is
represented in simplified form as say-v1 sem-str
uc 1 2 inform agent
var1 theme var2 Note varX
means the meaning of varX in the syn-struc.
9
TMR Structure Graphic
Author_Event_1
Theme
Inform_1
Agent
Theme
Decrease_1
Dresser Industries
Instrument
Theme
Expend_1
Import_1
Theme
Theme
Source
Destination
Japan
USA
Money_1
??
Purpose
Increase_1
Theme
Manufacture_1
Theme
10
Simplified Text Meaning Representation 1
Dresser Industries said it expects that major
capital expenditure for expansion of U.S.
manufacturing capacity will reduce imports from
Japan is simplified TMR author-event-1 agent va
lue unknown theme value inform-1 time time-be
gin inform-1.time-end time-end unknown
11
Simplified Text Meaning Representation 2
inform-1 agent value Dresser Industries
theme value decrease-1 time time-begin unkno
wn time-end (import-1.time-begin) ( (increase-1.time-begin)
12
Simplified Text Meaning Representation 3
decrease-1 agent value unknown theme
value import-1 instrument value expend-1 time
time-begin ( inform-1.time-end) (
expend-1.time-begin) ( import-1.time-begin)
time-end 13
Simplified Text Meaning Representation 4
import-1 agent value unknown theme
value unknown source value Japan
destination value USA time time-begin (
inform.time-end) (e-end unknown
14
Simplified Text Meaning Representation 5
expend-1 agent value unknown
theme value money-1
amount value 0.7 purpose
value increase-1 time time-begin
inform.time-end time-end e
15
Simplified Text Meaning Representation 6
increase-1 agent value unknown theme
value manufacture-1.theme time time-begin (
inform.time-end) ( time-end unknown manufacture-1 agent
value unknown theme
value unknown location value USA time time-b
egin inform.time-end time-end unknown
16
Simplified Text Meaning Representation 7
modality-1 type potential meaning of
expects value 1 maximum
value of scope decrease-1 potential modality
-2 type potential meaning of
capacity value 1 scope manufacture-1 co-re
ference-1 increase-1.agent
manufacture-1.agent co-reference-2
import-1.theme manufacture-1.theme
17
Ontological Semantics

• Based on computer understanding of the
  information
• Takes full advantage of the new technologies in
  computational semantics
• Problem-driven

18
Resources of Ontological Semantics
19
Input Text, Query,etc.
Output Text, Filled Template, Query Answer,
etc.
InputTextAnalyzer
Output Generator IE, QA, MT Engines, etc.
Application-Oriented InferenceEngine
Language-Dependent Static Knowledge Sources
Non-SemanticKnowledge Ecology Morphology,
Syntax, etc.
20
What Does Ontological Semantics do?

• Ontological semantics is a theory of meaning in
  natural language and a practical approach to
  natural language processing (NLP) that uses a
  real node-property deep-meaning ontology as a
  central resource for
• extracting and representing meaning of natural
  language texts,
• reasoning about knowledge derived from texts,
  and, if required,
• generating texts from meaning representations

21
Advantages of Ontological Semantics

• Ontological semantics is characterized by
• emphasis on knowledge content rather than
  format
• task driven rather than method driven
  approach
• variable grain size of description, e.g. -
  selectional restrictions for NLP - complex
  events for planning, etc. - physical
  properties for robotic vision

22
Some Basic Problems in NLP Ambiguity

• Syntactic ambiguity
• Japanese push bottles up Chinese
• Word Sense Ambiguity
• The coach lost a set
• World knowledge needed for disambiguation The
  soldiers shot at the women and I saw some of them
  fall

23
(No Transcript)
24
Non-Monotonic Inheritance
Triple Inheritance hierarchy for Nation
25
Lexical Entry for (the Nation of) Turkey
26
Text Meaning Representation

• proposition _1
• head exit_1
• agent human_54 Mr. Smith
• source location_23 London
• destination location_25 Ankara
• means vehicle_65 Boeing 757
• tmr-time
• time-begin YYYYMMDD July 2, 2000
• aspect
• iteration single phase end departed
• polarity positive
• mood indicative

27
Formalisms and Architectures

• Includes format and semantics of
• text meaning representations (TMRs)
• ontology
• fact and proper name DBs
• lexicons
• Generic processing architecture for
• analysis of meaning
• manipulation of meaning, including generation of
  text off of it

28
Description in Ontological Semantics

• Includes content of all knowledge sources, both
  static and dynamic
• implemented to provide full coverage for a
  language (or languages)
• In practice, ontological semantic description is
  always partial
• covering a subset of domains and sublanguages
• constantly under development
• through the process of acquisition
• and as modified by the operation of any
  applications based on the acquired descriptions

29
Methodology in Ontological Semantics

• Description Methodology
• Acquisition of static knowledge sources
• What are attainable levels of automation?
• What are specific techniques for acquisition of
  ontology? (see Nirenburg and Raskin 2004,
  Triezenberg 2005, for acquisition of other
  knowledge sources)
• Application Methodology
• TMRs may be extended in a well-defined way to
  support a specific application
• May require modifying the static resources and
  procedures used by the application

30
How to Tell a Real Ontology from a 5 Rolex
Ontology?

• Nodes with interrelated properties
• Properties part of ontology
• Non-monotonic inheritance
• Focus on content, not formalism
• Acquisition toolbox hybrid semi-automatic
• Domain extension capability
• Fully automatic applications

31
Legacy Resources

• An ontology of 5,500 concepts (80,000
  property-value pairs)
• Ontological-semantic lexicons English, 20,000
  entries (Spanish, 40,000 entries Chinese,
  2,500 entries others)
• Ecological preprocessor (punctuation, print
  conventions, morphology, syntax)--general-domain
  working version
• An ontological-semantic analyzer (text ?
  TMR)--general-domain pilot version
• An ontological-semantic generator (TMR ?
  text)--general-domain proof-of-concept

32
Plan of Presentation

• A Snapshot of Ontological Semantics
• Computational Humor
• Ontology of Emotion (by Katrina
• Triezenberg)
• Ontology of Persuasion
• Humor in Persuasion

33
Computer Implementation of GTVH

• Underlying theory of humor
• Goals of computation of humor
• Ontological semantic approach to humor
  computation

34
SSTH Main Hypothesis

• Joke-bearing text is compatible, in full or in
  part, with two different scripts
• The two scripts compatible with a joke- bearing
  text are opposite in a predefined way (a small
  list of oppositions

35
SSTH Script Oppositions

• If Script (Scenario) 1 is then Script (Scenario)
  2 is
• actual non-actual
• normal abnormal
• possible impossible
• good bad
• life death
• sexual non-sexual
• money-related non-money-related
• high-stature low-stature

36
Famous Polish/Light-Bulb Joke

• How many Polaks does it take to change a light
  bulb?Five. One to hold the bulb and four to turn
  the table hes standing on.

37
Joke Generation Lite

• Template
• how many lexical entry head does it take to
  change a light bulb? number_1 -number_2 to
  activity_1 and number_2 to activity_2.
• Filler
• head Polish Americans (Poles, Polaks)
• Trait dumb
• activity_1 hold light bulb
• number_1 5
• activity_2 turn table
• number_2 4

38
Computational Humor Dichotomies

• Switch modes from bona-fide to joke-telling or
  not?
• Build a joke generation system on top of a
  full-fledged meaning-based NLP system or not?
• Build a joke generation system based on the
  formal manipulation of text elements without any
  understanding

39
GTVH Knowledge Resources of the Joke
40
KRs of the Polish/Light-Bulb Joke

• LA How many Poles does it take to screw in a
  light bulb? Five. One to hold the light bulb and
  four to turn the table he's standing on.
• NS Question/answer
• TA Polish Americans
• SI changing a light bulb
• LM ground-figure reversal
• SO type normal/abnormal subtype smart/dumb

41
Where Is Computational Humor Now?

• the need for basing computational humor on a
  well-defined formal rigorous theory and the
  script-based semantic theory of humor (SSTH
  Raskin 1985) and its extension/revision, the
  general theory of verbal humor (GTVH Attardo and
  Raskin 1991, Attardo 1994), as the prime
  candidates for this role
• the feasibility of computational humor before the
  general problem of natural language processing
  (NLP), including full meaning representation, has
  been largely solved
• the ontological-semantic basis for all work in
  computational humor.

42
What Is Computational Humor For?

• The mature Stock hypothesis is well expressed in
  the advertisement for this Workshop on its Web
  site. The following applications are listed
  there
• business world applications (such as
  advertising, e-commerce, etc...)
• general computer-mediated communication and
  human-computer interaction
• increase the friendliness of natural language
  interfaces
• edutainment and autonomous agents systems.

43
What Is Computational Humor For? (contd)

• Add two more types of applications
• customer acceptance enhancement (for instance in
  IAS, where sysadmins sabotage the installation of
  security software--so entertain the bastards!)
• humor detection as part of the semantic forensics
  enterprise, where deception, omission, etc., are
  automatically detected

44
Plutchiks Emotional Solid
45
Ontology of Emotion

• 45 new and recycled concepts, described by
• Taking advantage of multiple inheritance
• Physiological responses
• Facial responses
• Stereotyped responses

46
Example of an Emotion Concept

• (LOVE
• (AGENT (SEM (HUMAN)))
• (THEME (SEM (OBJECT)))
• (IS-A (VALUE (happiness trust)))
• (subclasses (value (appreciate)))
• (DEFINITION (VALUE ("to feel strong affection
  for")))
• (has-event-as-part (sem (protect help
  interact-socially)))
• (intensity (sem (.5))))

47
Example of an Emotion Concept

• (humor
• (definition (value (to find something funny)))
• (is-a (value (emotional-event)))
• (has-event-as-part (sem (
• laugh
• smile-zygomatic
• parasympathetic-response
• )))
• )

48
Where to go

• Expand text to a much finer grainsize
• Expand concepts to a finer grainsize, writing
  full scripts where appropriate
• Expand domain to include motivation
• Create the lexicon, probably using a thesaurus
  rather than a corpus

49
Plan of Presentation

• A Snapshot of Ontological Semantics
• Computational Humor
• Ontology of Emotion (by Katrina
• Triezenberg)
• Ontology of Persuasion
• Humor in Persuasion

50
Ontology of Persuasion

• Two Different Aspects
• Representation of persuasion in ontological
  semantics
• Computation recognition of persuation and of its
  goals

51
Ontological Representation of Persuasion

• No problem
• Volitive and epistemic modality the persuader
  wants (volitive modality value of 1) an event,
  whose agent is the persuadee, to become reality
  (epistemic modality value of 1)
• Property of effect that event is the effect of
  the event of which the persuader is the agent

52
Ontological Representation of Persuasion (contd)

• Or in English
• Persuader The Yugo is the best car in the world
• Persuader wants the persuadee to buy a Yugo (from
  the persuaders mistresss husband?)
• Persuadee goes and buys a Yugo

53
Ontological Semantic Computation of Persuasion

• Persuader The Yugo is the best car in the world
• A simple act of stupidity or persuasion?
• Challenge collecting the clues, the way humans
  do, to guess the goal
• Trivial clues
• Assertion persuasion to believe?
• Praise persuasion to appropriate (buy a
  commodity, spouse, faith)?
• Criticism etc.?

54
Plan of Presentation

• A Snapshot of Ontological Semantics
• Computational Humor
• Ontology of Emotion (by Katrina
• Triezenberg)
• Ontology of Persuasion
• Humor in Persuasion

55
Humor and Persuasion

• Commonly Held Views
• Humor persuades by improving the ambiance
• Humor makes the persuader more likable and the
  persuadee more malleable
• Humor incapacitates and manipulates the persuadee
• Humor distracts the persuadee
• Humor eliminates the stress of the decision

56
Humor and Persuasion (contd)

• More interesting and totally unexplored
• Humor persuades by offering a likable logical
  parallel
• Persuader wants persuadee to reject a third
  partys explanation of an important prior event
• Persuader retells the alligator and chimp joke
• Persuadee accepts the explanation as equally
  implausible
• And now for the last slide--with the joke!

57
Persuasion With a Joke

• The alligator and chimp joke (NYC, circa 1930)
• The Big Depression. The vaudeville world. This
  performer seeks employment. His trained alligator
  and chimp perform a musical number the chimp
  sings while the alligator accompanies her on the
  piano. Smashing success. Reporters stalk the
  performer trying to get the secret. Properly
  boozed up, he confesses the alligator both plays
  the piano and sings--the chimp just lip-syncs!

58
Hey, that was not the last slide!--Inconclusive
Conclusion

• More questions raised than answers provided!
• Interesting work to do--both within
  linguistics/NLP/ontological semantics and in
  almost all the disciplines represented at this
  Workshop
• Thank you!
• Questions?

59
Oy, gevalt--more slides!
60
No, not really
61
(No Transcript)