Metaphor comprehension
This presentation is the property of its rightful owner.
Sponsored Links
1 / 21

Metaphor Comprehension PowerPoint PPT Presentation


  • 213 Views
  • Uploaded on
  • Presentation posted in: General

Metaphor Comprehension. Presenter : Quamrul Islam Khan 74.793 Natural Language and Speech Processing. Outline. Metaphor Metaphor & Analogy Available Systems SME (analogy reasoning) Sapper (metaphor interpretation). What is Metaphor?. From the semiotic point of view : “a metaphor is a

Download Presentation

Metaphor Comprehension

An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -

Presentation Transcript


Metaphor comprehension

Metaphor Comprehension

Presenter : Quamrul Islam Khan

74.793 Natural Language and Speech Processing


Outline

Outline

  • Metaphor

  • Metaphor & Analogy

  • Available Systems

    • SME (analogy reasoning)

    • Sapper (metaphor interpretation)


What is metaphor

What is Metaphor?

From the semiotic point of view : “a metaphor is a

dynamic, as opposed to stable, sign, and this

follows the etymology of the word, which suggests

a transfer or displacement of names”.

(Veale, 1995)


What is metaphor contd

What is Metaphor?(contd.)

  • Metaphor can be described “as the act or process of denoting one concept (X) with a sign conventionally tied to another concept (Y), with the purpose of

  • (i) emphasizing certain associations of the (X) over others (e.g. my dentist is a barbarian);

  • (ii)enriching the conceptual structure of the (X) by analogy with another domain (the CPU is the brain of the computer);

  • (iii) conveying some aspect of the (X) which defies conventional lexicalization (the leg of the chair, the neck of the bottle)”

    (Veale, 1995)


What is analogy

What is Analogy?

  • An analogy is the

    • mapping of knowledge from one domain to another

    • Conveys a system of relations between those domains


Metaphor analogy contd

Metaphor & Analogy (contd.)

  • Analogy and analogical reasoning can comprehend metaphors.

  • Some of the systems on analogy and metaphors

    • Structure-Mapping Engine (SME) is the program that can perform analogical processing.

    • Sapper is a hybrid model for metaphor interpretation

    • Analogical Constraint Mapping Engine (ACME) is an analogical reasoning system.


Structure mapping engine sme

Structure Mapping Engine (SME)

  • “An analogy is the mapping of knowledge of one domain (base) into another domain (target) which convey that a system of relations which holds in (base) also holds in (target)” (Falkenhainer et al., 1989).

  • It is the structural properties that determine the content of analogy.

    • The inter-relationships between the facts of the base and the target domain


Overview on sme

Overview on SME

  • SME takes two propositional descriptions as input, which are the base and the target.

  • The output gmap, which is an interpretation of the comparison of the base and target contains

    • Correspondences (linking items in the base and the target)

    • A set of candidate inference (statement in base can be inferred to holds in target based on the information of the correspondences that are formed)

    • A structural evaluation score (provides the match quality)


Sme phases

SME Phases

  • Three analogical processing phase.

    • Access: retrieval of a base description given a target situation

    • Mapping and Inference: mapping is the correspondence between the base and target. A mapping with additional knowledge in the base can be transferred to the target is the candidate inference of the analogy.

    • Evaluation and Use: estimate the quality of the match


Structure mapping engine contd

Structure Mapping Engine (contd.)

  • How to estimate quality

    • Structural: (1) the number of similarities and differences, (2) the degree of similarity and difference, and (3) amount and type of new knowledge the analogy provides through the mapped candidate inference

    • Validity: the generated inference must be checked using the available world knowledge

    • Relevance: check whether the resultant analogy is useful to the reasoner’s purpose


Example of sme

Example of SME

Base domain:

1 (IMPLIES 2 (AND 3 (SENSITIVE-TO 4 LITMUS32 5 ALCOHOL-VAPOUR) 6 (INSIDE 7 COOLANT 8 SUMP)

10 (HELD-CLOSE LITMUS32 SUMP) )

11 (DETECTABLE 12 (GIVES-OFF COLLANT ALCOHOL-VAPOUR)))

13 (IMPLIES 14 (LIQUID COOLANT) 15 (POSSIBLE (GIVES-OFF COOLANT ALCOHOL-VAPOUR)))

16 (IMPLIES 17(DECREASED 19 (PRESSURE SUMP))

20 (INCREASED 21(FLOW-RATE 22 (FLOW 23 STILL SUMP COOLANT 24 PIPE))))

26 (IMPLIES 27(INCREASED (PRESSURE SUMP))

28 (DECREASED (FLOW-RATE (FLOW STILL SUMP COOLANT PIPE))))

29 (IMPLIES 30 (DECREASED 31(AREA PIPE)) (DECREASED (FLOW-RATE (FLOW STILL SUMP COOLANT PIPE))))

32 (IMPLIES 33 (INCREASED (AREA PIPE)) (INCREASED (FLOW-RATE (FLOW STILL SUMP COOLANT PIPE))))

33 (CAUSE 34 (GREATER 35 (PRESSURE STILL) (PRESSURE SUMP)) (FLOW STILL SUMP COOLANT PIPE))

36 (FLAT-TOP COOLANT)

Target domain:

1 (INCREASED 2 (FLOW-RATE 3 (FLOW 4 EFFLUENT 5 HEAT-SINK 6 HEAT 7 HX)))

8 (DECTECTABLE 9 (GIVES-OFF EFFLUENT 10 RADIATION))

11 (CUASE 12 (CONTAINS EFFLUENT 13 STRONGTIUM0) (GIVES-OFF EFFLUENT RADIATION))

14 (LIDUID EFFLUENT)

15 (FLAT-TOP EFFLUENT)

16 (GREATER 17 (TEMPERATURE EFFLUENT) 18(TEMPERATURE HEAT-SINK))

From Forbus (1990)


Example of sme1

Example of SME

From Forbus (1990)


Drawbacks of sme

Drawbacks of SME

  • SME constructsall the structurally consistent interpretations of an analogy which make the algorithm computationally inefficient

  • SME failsto generate the output analogy which is useful for the reasoner’s purpose


Improvement of sme

Improvement of SME

  • Greedy merge algorithm

    does the gmap construction where the pmap with best score is selected to be combined and form the new gmap

  • Pragmatic marking

    identifies interpretations which are relevant


Sapper

Sapper

  • Views the interpretation of metaphor as connectionist bridge building

  • Uses the existing structure of semantic memory to interpret different concepts

  • Represents concepts as nodes and arcs between the nodes as the semantic relation between these concepts

  • Bottom-up approach to comprehend metaphor


Sapper contd

Sapper (contd.)

  • If Butcher is like Surgeon

  • Then Abattoir is like Operating-Theatre

  • and Meat is like Human-flesh

  • and Cleaver is like Scalpel

  • and Carcass is like Corpse

  • and Slaughter is like Surgery

  • These hypothesis are driven by

  • literal similarity (both Cleavers and Scalpels are sharp and metallic)

  • Higher-order similarity (the relations between them can be drawn like Cleaver supports Slaughtering and Scalpel supports Surgery)

From Veale (1995)


Sapper contd1

Sapper (contd.)

  • The Triangulation rule is applied in the nodes which share common association

  • The Squaring rule is applied to the formed triangulation and forms bridges between nodes

From Veale (1995)


Comparative analysis

Comparative Analysis

From Veale (1995)


Recent research on metaphor

Recent Research on Metaphor

  • “A Statistical Approach to Metaphor

    Processing” a proposed system by Julia

    Birke and Anoop Sarkar of Simon Fraser

    University.

  • They are working on Example-based

    Machine Translation system to comprehend

    metaphor


Conclusion

Conclusion

  • SME and Sapper are competent for predicate-centered representation

  • Sapper gives better interpretation for object-centered representatione.g. Surgeon is like a butcher.


References

References

Falkenhainer, B., Forbus, K. D., and Genter, D.. The Structure-Mapping Engine: Algorithm and Examples, Artificial Intelligence. Vol. 41, pp. 1-63, 1989.

Ferguson, R. W., Forbus, K. D., & Gentner, D. (1997). On the proper treatment of noun-noun metaphor: A critique of the Sapper model. Poster presented at the 19th Annual Meeting of the Cognitive Science Society.

Forbus, K. D., and Oblinger, D.. Making SME Greedy and Pragmatic, Proceeding of CogSci’ 90, 1990.

Veale, T. “Metaphor, Memory and Meaning: Symbolic and Connectionist Issues in Metaphor Interpretation”, School of Computer Application, Dublin City University, 1995.http://www.compapp.dcu.ie/~tonyv/

Veale, T., and Keane, M.. Epistemological Pitfalls in Metaphor Comprehension: A Comparison of Three Models and a New Theory of Metaphor. The International Cognitive Linguistics Conference, ICLA’95, 1995.

Birke, J. and Sarkar, A.. “A Statistical Approach to Metaphor Processing”

http://css.sfu.ca/sites/natlang/researchProject.php?s=84


  • Login