Computational Approaches to Creative Language: Summary Caroline - - PowerPoint PPT Presentation

computational approaches to creative language summary
SMART_READER_LITE
LIVE PREVIEW

Computational Approaches to Creative Language: Summary Caroline - - PowerPoint PPT Presentation

Computational Approaches to Creative Language: Summary Caroline Sporleder Computational Linguistics Universit at des Saarlandes Sommersemester 2010 20.07.2010 Caroline Sporleder Computational Approaches to Creative Language Wrapping-Up


slide-1
SLIDE 1

Computational Approaches to Creative Language: Summary

Caroline Sporleder

Computational Linguistics Universit¨ at des Saarlandes

Sommersemester 2010 20.07.2010

Caroline Sporleder Computational Approaches to Creative Language

slide-2
SLIDE 2

Wrapping-Up

Caroline Sporleder Computational Approaches to Creative Language

slide-3
SLIDE 3

Creativity (Boden (1990, 1999))

Creativity The generation of ideas that are novel and valuable. Value is typically negotiated by social groups. AI models of creativity combinatorial creativity exploration/transformation of conceptual space

exploratory creativity transformational creativity

Caroline Sporleder Computational Approaches to Creative Language

slide-4
SLIDE 4

Creative Language: Overview

Detection & Processing idioms (type-based, token-based) metaphor neologisms irony, sarcasm humor Generation metaphor humor poetry stories

Caroline Sporleder Computational Approaches to Creative Language

slide-5
SLIDE 5

Detection & Analysis: Techniques (1)

Idioms Metaphor

Caroline Sporleder Computational Approaches to Creative Language

slide-6
SLIDE 6

Detection & Analysis: Techniques (1)

Idioms type-based

association measures (idiom components) (Villavicencio et al., 2007) statistical measures of linguistic fixedness (Bannard, 2007)

token-based

cohesion with context (Li & Sporleder, 2009) vector space models (Katz & Giesbrecht, 2006)

Metaphor

Caroline Sporleder Computational Approaches to Creative Language

slide-7
SLIDE 7

Detection & Analysis: Techniques (1)

Idioms type-based

association measures (idiom components) (Villavicencio et al., 2007) statistical measures of linguistic fixedness (Bannard, 2007)

token-based

cohesion with context (Li & Sporleder, 2009) vector space models (Katz & Giesbrecht, 2006)

Metaphor knowledge-based systems

met* (literal, metonymy, metaphor, anomaly) (Fass, 1991) MIDAS (source target mapping; metaphor maps; new metaphors via analogy) (Martin, 1992) slipnets (Veale & Hao, 2008)

knowledge-lean systems for metaphor detection (similar to idiom detection) (Birke & Sarkar, 2006)

Caroline Sporleder Computational Approaches to Creative Language

slide-8
SLIDE 8

Detection & Analysis: Techniques (2)

Neologisms Irony, Sarcasm Humor

Caroline Sporleder Computational Approaches to Creative Language

slide-9
SLIDE 9

Detection & Analysis: Techniques (2)

Neologisms Zeitgeist (Wikipedia / WordNet, plus analysis rules) (Veale & Butnario, 2010) statistical methods (Cook & Stevenson, 2010) Irony, Sarcasm Humor

Caroline Sporleder Computational Approaches to Creative Language

slide-10
SLIDE 10

Detection & Analysis: Techniques (2)

Neologisms Zeitgeist (Wikipedia / WordNet, plus analysis rules) (Veale & Butnario, 2010) statistical methods (Cook & Stevenson, 2010) Irony, Sarcasm bag-of-words, surface cues (Burfoot & Baldwin, 2009; Carvalho et al., 2009; Kreuz & Caucci, 2007) pattern, n-gram based (Tsur et al., 2010) Humor

Caroline Sporleder Computational Approaches to Creative Language

slide-11
SLIDE 11

Detection & Analysis: Techniques (2)

Neologisms Zeitgeist (Wikipedia / WordNet, plus analysis rules) (Veale & Butnario, 2010) statistical methods (Cook & Stevenson, 2010) Irony, Sarcasm bag-of-words, surface cues (Burfoot & Baldwin, 2009; Carvalho et al., 2009; Kreuz & Caucci, 2007) pattern, n-gram based (Tsur et al., 2010) Humor classification based on phonological, stylistic, lexical properties (Mihalcea & Strapparava, 2005)

Caroline Sporleder Computational Approaches to Creative Language

slide-12
SLIDE 12

Generation: Techniques

Metaphor Humor Poetry Stories

Caroline Sporleder Computational Approaches to Creative Language

slide-13
SLIDE 13

Generation: Techniques

Metaphor web harvesting plus WordNet (Sardonicus, simile generation) (Veale & Hao, 2007) Humor Poetry Stories

Caroline Sporleder Computational Approaches to Creative Language

slide-14
SLIDE 14

Generation: Techniques

Metaphor web harvesting plus WordNet (Sardonicus, simile generation) (Veale & Hao, 2007) Humor knowledge-based pun generation using templates, phonological & taxonomic knowledge (Manurung et al., 2008) Poetry Stories

Caroline Sporleder Computational Approaches to Creative Language

slide-15
SLIDE 15

Generation: Techniques

Metaphor web harvesting plus WordNet (Sardonicus, simile generation) (Veale & Hao, 2007) Humor knowledge-based pun generation using templates, phonological & taxonomic knowledge (Manurung et al., 2008) Poetry Haiku generation using word association norms and structure templates (Netzer et al., 2009) Stories

Caroline Sporleder Computational Approaches to Creative Language

slide-16
SLIDE 16

Generation: Techniques

Metaphor web harvesting plus WordNet (Sardonicus, simile generation) (Veale & Hao, 2007) Humor knowledge-based pun generation using templates, phonological & taxonomic knowledge (Manurung et al., 2008) Poetry Haiku generation using word association norms and structure templates (Netzer et al., 2009) Stories knowledge-based, e.g. using case-based reasoning (Gerv´ as et al., 2004) knowledge-lean, schemas harvested from existing stories and re-combined (McIntyre & Lapata, 2010)

Caroline Sporleder Computational Approaches to Creative Language

slide-17
SLIDE 17

Summary (1)

Techniques knowledge-based (KBs and semantic networks; metaphors, story generation) less knowledge-based (templates, taxonomies; poetry and pun generation) using internet resources (Wikipedia; neologisms, simile generation) data-driven

supervised ML using shallow features (detection of metaphors, irony, sarcasm, humor) corpus-based, statistical, unsupervised (type- and token-based idiom detection, neologisms, story generation)

Caroline Sporleder Computational Approaches to Creative Language

slide-18
SLIDE 18

Summary (2)

Types of Creativity involved combinatorial / exploratory (web corpus-based re-combination, implicit conceptual space)

Veale & Hao (2007) - simile generation McIntyre & Lapata (2010) - story generation

exploratory (explicitly defined conceptual space)

Gerv´ as et al. (2004) - story generation Manurung et al. (2008) - pun generation ???Netzer et al. (2009) - Haiku generation

transformational

??? does this make sense for machines ???

Caroline Sporleder Computational Approaches to Creative Language

slide-19
SLIDE 19

What does a computer have to be able to do?

detect? understand? generate? idioms metaphors humor irony poetry stories

Caroline Sporleder Computational Approaches to Creative Language

slide-20
SLIDE 20

What does a computer have to be able to do?

detect? understand? generate? idioms definitely definitely possibly metaphors humor irony poetry stories

Caroline Sporleder Computational Approaches to Creative Language

slide-21
SLIDE 21

What does a computer have to be able to do?

detect? understand? generate? idioms definitely definitely possibly metaphors definitely definitely posibly humor irony poetry stories

Caroline Sporleder Computational Approaches to Creative Language

slide-22
SLIDE 22

What does a computer have to be able to do?

detect? understand? generate? idioms definitely definitely possibly metaphors definitely definitely posibly humor probably possibly possibly irony poetry stories

Caroline Sporleder Computational Approaches to Creative Language

slide-23
SLIDE 23

What does a computer have to be able to do?

detect? understand? generate? idioms definitely definitely possibly metaphors definitely definitely posibly humor probably possibly possibly irony definitely definitely possibly poetry stories

Caroline Sporleder Computational Approaches to Creative Language

slide-24
SLIDE 24

What does a computer have to be able to do?

detect? understand? generate? idioms definitely definitely possibly metaphors definitely definitely posibly humor probably possibly possibly irony definitely definitely possibly poetry possibly ??? possibly stories

Caroline Sporleder Computational Approaches to Creative Language

slide-25
SLIDE 25

What does a computer have to be able to do?

detect? understand? generate? idioms definitely definitely possibly metaphors definitely definitely posibly humor probably possibly possibly irony definitely definitely possibly poetry possibly ??? possibly stories — possibly possibly

Caroline Sporleder Computational Approaches to Creative Language

slide-26
SLIDE 26

What does a computer have to be able to do?

detect? understand? generate? idioms definitely definitely possibly metaphors definitely definitely posibly humor probably possibly possibly irony definitely definitely possibly poetry possibly ??? possibly stories — possibly possibly Challenge: development of relatively knowledge-lean methods for these tasks!

Caroline Sporleder Computational Approaches to Creative Language

slide-27
SLIDE 27

Reading Tips from ACL 2010

Ekaterina Shutova. “Models of Metaphor in NLP”. ACL 2010 Beata Beigman Klebanov, Eyal Beigman. “A Game-Theoretic Model of Metaphorical Bargaining”. ACL 2010 Neil McIntyre, Mirella Lapata. “Plot Induction and Evolutionary Search for Story Generation”. ACL 2010

Caroline Sporleder Computational Approaches to Creative Language

slide-28
SLIDE 28

Term Paper

Caroline Sporleder Computational Approaches to Creative Language

slide-29
SLIDE 29

Term Papers

Writing term papers answer a concrete research question (or a set of related questions) show evidence of independent thinking and reasoning independent bibliographic work meet academic standards (proper citations/references, well-structured, well-written etc.) Don’t plagiarise! can include some practical work but doesn’t have to around 100 hours of work (2-3 weeks full-time) 15-20 pages long 10-15 references Please discuss topic with me beforehand!

Caroline Sporleder Computational Approaches to Creative Language