SlideShare uma empresa Scribd logo
1 de 40
Arguing from a Point of View

           Adam Wyner1 and Jodi Schneider2
     1 - Department of Computer Science, University of Liverpool
2 – Digital Enterprise Research Institute, National University of Ireland

      First International Conference on Agreement Technologies
     Centre for Advanced Academic Studies, University of Zagreb
                           Dubrovnik, Croatia

                           October 16, 2012
Overview
• Hotel reviews are a source of arguments.
• Point of view is needed to evaluate arguments such as
      – The hotel is in an excellent location.
• Therefore we relativise evaluative statements based
  on point of view.
• The key point: evaluative statements can be justified
  using instantiated argumentation schemes relative to
  a user and a domain model.


October 16, 2012            Wyner and Schneider, AT 2012   1
Hotel use case




October 16, 2012     Wyner and Schneider, AT 2012   2
Positive reviews




October 16, 2012      Wyner and Schneider, AT 2012   3
Negative reviews




October 16, 2012      Wyner and Schneider, AT 2012   4
TMI




• How much 'bad' spoils what amount of 'good'?
• How do the scores relate to the content? How does the
  content justify or argue for the score given?
• How do the comments relate to one another? Linear text &
  lists of comments aren’t rich enough: Elaborate network of
  point and counterpoint.
October 16, 2012        Wyner and Schneider, AT 2012           5
It's all about YOU!
People don't just want “information”
They want information that is
• relevant to them
• appeals to them
• sees things from their point of view.




October 16, 2012       Wyner and Schneider, AT 2012   6
Goal
• To support relativised argumentation derived from
  distributed, inconsistent information.




October 16, 2012    Wyner and Schneider, AT 2012      7
Evaluative expressions use case
                          - client and travel agent -
          I'm going to a conference in venue X in Valencia and need a hotel
          room.
  Bill

          Hotel Valencia is in an excellent location.
Travel agent
          Why do you say it is an excellent location?

          The hotel is a kilometer from the venue X. And the hotel is in the
          old part of the city.

          OK, please book it.

    October 16, 2012              Wyner and Schneider, AT 2012                 8
Evaluative expressions use case
                       - client and travel agent -
       I'm going to a conference in venue X in Valencia and need a hotel
       room.
Jill

       Hotel Valencia is in an excellent location.

       Why do you say it is an excellent location?

       The hotel is a kilometer from the venue X. And the hotel is in the
       old part of the city.

       But it is a noisy and trashy old part. And it is too far. Please find me
       something else.
 October 16, 2012              Wyner and Schneider, AT 2012                   9
Argument evaluation is user-relative
• Bill & Jill receive the same argument from the travel
  agent but evaluate it differently.
• Given the premises
   – The hotel is a kilometer from the venue X. And the
      hotel is in the old part of the city.
•     Bill has accepted the claim
   – Hotel Valencia is in an excellent location.
•      Given the same premises, Jill has not accepted
       the claim (and doesn't even agree with all the
  premises).
• Different ways to argue for and against the same claim.
October 16, 2012      Wyner and Schneider, AT 2012          10
Approach
• Argumentation schemes are key
      – Normative patterns of defeasible reasoning.
      – Variables can be seen as targets for information
        extraction. Could use text analysis to instantiate.
      – Evaluate instantiated arguments using
        argumentation frameworks.
• Relativise the instantiated arguments to a
  user.

October 16, 2012         Wyner and Schneider, AT 2012         11
Argumentation Schemes Overview
• Example scheme from the literature “Credible
  source”:
   – Instantiated
   – Abstract
   – Questions used to critique the argument
• Two new schemes for our use case
   – “Evaluation of location”
   – “Evaluation of quality”


October 16, 2012    Wyner and Schneider, AT 2012   12
Argumentation scheme example
                      - instantiated -
• Normative patterns of defeasible reasoning:
   – Dr. Rose is an expert about road safety;
   – Dr. Rose asserts that having more speed cameras will save
     more lives;
   – Having more speed cameras will save lives is a statement
     concerning road safety;
   – Dr. Rose is credible about road safety;
   – and Dr. Rose is reliable;
   – Therefore, it is presumably true that having more speed
     cameras will save more lives.


October 16, 2012        Wyner and Schneider, AT 2012             13
Argumentation scheme example
                       - abstracted -
• Normative patterns of defeasible reasoning:
   – X is an expert about Y;
   – X asserts Z;
   – Z is a statement concerning Y;
   – X is credible about Y;
   – and X is reliable;
   – therefore, it is presumably true that Z.




October 16, 2012        Wyner and Schneider, AT 2012   14
Argumentation scheme example
                        - critique -
• Questions used to critique the argument:
   – How credible is X as an expert source?
   – Is the claim about Z consistent with what other
     experts assert?
   – Is X’s assertion based on evidence?
   – Others....




October 16, 2012        Wyner and Schneider, AT 2012   15
Use case elements
• New argumentation schemes:
   – Evaluation of location.
   – Evaluation of quality.
• Instantiate schemes relative to a user model.
• Domain and evaluative terminology.
• User model – selection from domain terminology plus
  some terminology for parameters, context of
  use, constraints....



October 16, 2012       Wyner and Schneider, AT 2012     16
User Information




 In this paper, we represent user models by terminology and
 instantiated schemes.

 In other work, we add these components.



October 16, 2012        Wyner and Schneider, AT 2012          17
“Evaluation of location” arg. scheme
                      - abstract -




October 16, 2012     Wyner and Schneider, AT 2012   18
“Evaluation of location” arg. scheme
             - Instantiating for Bill & Jill -




October 16, 2012      Wyner and Schneider, AT 2012   19
“Evaluation of quality” arg. scheme
                        -abstract-




October 16, 2012       Wyner and Schneider, AT 2012   20
“Evaluation of quality” arg. scheme
                  -Instantiating for Bill-




October 16, 2012       Wyner and Schneider, AT 2012   21
Instantiating for Jill




October 16, 2012        Wyner and Schneider, AT 2012   22
Use case elements
• Argumentation schemes:
   – Evaluation of location.
   – Evaluation of quality.
• Instantiate schemes relative to a user model.
• Domain and evaluative terminology.
• User model – selection from terminology and
  instantiated schemes.



October 16, 2012       Wyner and Schneider, AT 2012   23
Domain and evaluative terminology




October 16, 2012   Wyner and Schneider, AT 2012   24
User-associated inference
• If the instantiations of both argumentation schemes
  are acceptable to the user, then the user has a
  justification to book the hotel.
• For us, the model of the user can be given in terms of
  a logical language – the terminology and the
  schemes instantiated with that terminology.
• Arguing about the instantiations, e.g. Jill's criticism of
  the travel agent's proposition, is a meta-argument
  about the contents of the user model.

October 16, 2012           Wyner and Schneider, AT 2012   25
Argumentation frameworks & text analysis

• This paper is part of a larger work on the argumentation
  pipeline, from textual source to abstract argumentation.
• Introduces new schemes and instantiates them relative
  to a user.
• Other parts:
   – We have a text analytic tool (GATE) to support the
     extraction of relevant information from the source.
   – We have a proposal for integrating this with
     argumentation frameworks.


October 16, 2012      Wyner and Schneider, AT 2012           26
Argumentation pipeline
        Source Text

                                                       Instantiated                           Abstract Argumentation
No fresh orange juice at                          Argumentation Schemes                             Frameworks
breakfast and besides terrible
filter coffee extra payment for
cappuchino etc... No wifi in the
rooms (says so in
description, but still...).                       AS1: ....
                                                                                 Relate
                                   Extract text
                                                                                 schemes to
Very impressive hotel with         to schemes                                    arguments.
stunning views. Staff were
attentive - especially the bell
boys. 5 min bus journey to the                    AS2: ....
old town or 15 min walk. The
room was very comfortable.


If u want to stay with comfort I
would never recommend this
hotel on arrival I was waiting                    AS3: ....
my room from 14.00 till
16.00, but again they gave me a
room with two separate beds
ignoring my comments in the
booking (one king bed and big
bathtube)


     October 16, 2012                             Wyner and Schneider, AT 2012                               27
Consumer argumentation scheme
   Variables in schemes as targets for extraction.

   Premises:
   • Camera X has property P.
   • Property P promotes value V for agent A.

   Conclusion:
   • Agent A should Action Camera X.

October 16, 2012       Wyner and Schneider, AT 2012   28
Identifying and extracting text
• Annotate text:
   – Simple or complex annotations.
   – Highlight annotations with
   – Search for and extract text by annotation.
• GATE “General Architecture for Text Engineering”.
   – Works with large corpora of text.
   – Rule-based or machine-learning approaches.



 October 16, 2012       Wyner and Schneider, AT 2012   29
,
                                                  ,




October 16, 2012   Wyner and Schneider, AT 2012       30
Query for patterns




October 16, 2012       Wyner and Schneider, AT 2012   31
An argument for buying the camera
Premises:
   The pictures are perfectly exposed.
   The pictures are well-focused.
   No camera shake.
   Good video quality.
   Each of these properties promotes image quality.

Conclusion:
  (You, the reader,) should buy the CanonSX220.


October 16, 2012        Wyner and Schneider, AT 2012   32
An argument for NOT buying the
                   camera
Premises:
   The colour is poor when using the flash.
   The images are not crisp when using the flash.
   The flash causes a shadow.
   Each of these properties demotes image quality.

Conclusion:
  (You, the reader,) should not buy the CanonSX220.




October 16, 2012        Wyner and Schneider, AT 2012   33
Counterarguments to the premises of
             “Don’t buy”

             The colour is poor when using the flash.
             For good colour, use the colour setting, not the flash.

             The images are not crisp when using the flash.
             No need to use flash even in low light.

             The flash causes a shadow.
             There is a corrective video about the flash shadow.


October 16, 2012               Wyner and Schneider, AT 2012            34
Argumentation Frameworks
 • <Arguments, Relation>, where
   arguments are atomic nodes
   and the relation is attack.
 • Calculate the sets of nodes that
   are 'compatible'.
 • Articulate nodes with a logical
   language of literals and
   rules, where attack is
   contrariness between
                                                          Preferred Extension of the AF.
   expressions of the language.



October 16, 2012           Wyner and Schneider, AT 2012                                    35
Future work
•    User model formalisation and meta-argumentation.
•    Text analysis for this set of data.
•    Tool refinement.
•    Add ontology modules to the tool.
•    Multi-critierial argumentation – properties ascribed
     to the argument vs. premises of the argument.




October 16, 2012        Wyner and Schneider, AT 2012        36
Related Papers
• Wyner, van Engers, and Hunter (2010). "Working on the Argument
  Pipeline: Through Flow Issues between Natural Language
  Argument, Instantiated Arguments, and Argumentation
  Frameworks", Workshop on Computational Models of Natural
  Argument (CMNA).
• Wyner, Schneider, Atkinson, and Bench-Capon (2012). ''Semi-
  automated argumentation analysis of online product
  reviews'', Conference on Computational Models of Argument
  (COMMA).
• Schneider and Wyner (2012). ''Identifying consumers' arguments in
  text'', Workshop on Semantic Web and Information Extraction
  (SWAIE at EKAW).
• Schneider, Davis, and Wyner (2012). ''Dimensions of argumentation
  in social media'', Conference on Knowledge Engineering and
  Knowledge Management (EKAW).
October 16, 2012         Wyner and Schneider, AT 2012             37
Acknowledgements
 •     FP7-ICT-2009-4 Programme, IMPACT Project, Grant
       Agreement Number 247228.
 •     Science Foundation Ireland Grant No. SFI/08/CE/I1380 (Líon-
       2). Short-term Scientific Mission grant from COST Action
       IC0801 on Agreement Technologies. SFI Short Term Travel
       Fellowship.




October 16, 2012           Wyner and Schneider, AT 2012          38
Thanks for your attention!

• Questions?
• Contacts:

      – Adam Wyner                 adam@wyner.info
      – Jodi Schneider             jschneider@pobox.com




October 16, 2012           Wyner and Schneider, AT 2012   39

Mais conteúdo relacionado

Semelhante a Arguing from a Point of View

Identifying consumers’ arguments in text swaie at ekaw 2012 10-09
Identifying consumers’ arguments in text swaie at ekaw 2012 10-09Identifying consumers’ arguments in text swaie at ekaw 2012 10-09
Identifying consumers’ arguments in text swaie at ekaw 2012 10-09jodischneider
 
Semi-automated argumentation analysis of online product reviews--COMMA 2012-0...
Semi-automated argumentation analysis of online product reviews--COMMA 2012-0...Semi-automated argumentation analysis of online product reviews--COMMA 2012-0...
Semi-automated argumentation analysis of online product reviews--COMMA 2012-0...jodischneider
 
Identifying arguments for evaluation using an argument explorer - London Argu...
Identifying arguments for evaluation using an argument explorer - London Argu...Identifying arguments for evaluation using an argument explorer - London Argu...
Identifying arguments for evaluation using an argument explorer - London Argu...jodischneider
 
Kdd 2014 Tutorial - the recommender problem revisited
Kdd 2014 Tutorial -  the recommender problem revisitedKdd 2014 Tutorial -  the recommender problem revisited
Kdd 2014 Tutorial - the recommender problem revisitedXavier Amatriain
 
Interactive XAI for ODSC East 2023
Interactive XAI for ODSC East 2023Interactive XAI for ODSC East 2023
Interactive XAI for ODSC East 2023Meg Kurdziolek
 
A Multidomain and Multilingual Conceptual Data Model for Online Reviews Repre...
A Multidomain and Multilingual Conceptual Data Model for Online Reviews Repre...A Multidomain and Multilingual Conceptual Data Model for Online Reviews Repre...
A Multidomain and Multilingual Conceptual Data Model for Online Reviews Repre...Marcirio Chaves
 
Jumpstart: Introduction to Schema Design
Jumpstart: Introduction to Schema DesignJumpstart: Introduction to Schema Design
Jumpstart: Introduction to Schema DesignMongoDB
 
Opinion-based User Profile Modeling for Contextual Suggestions
Opinion-based User Profile Modeling for Contextual SuggestionsOpinion-based User Profile Modeling for Contextual Suggestions
Opinion-based User Profile Modeling for Contextual SuggestionsTwitter Inc.
 
Intro to R and Data Mining 2012 09 27
Intro to R and Data Mining 2012 09 27Intro to R and Data Mining 2012 09 27
Intro to R and Data Mining 2012 09 27Raj Kasarabada
 
World Usability Day 2014: Engagement (Master Deck)
World Usability Day 2014: Engagement (Master Deck)World Usability Day 2014: Engagement (Master Deck)
World Usability Day 2014: Engagement (Master Deck)NYCUXPA
 
How To Use Benchmarking To Evaluate Performance And Maximize Revenue
How To Use Benchmarking To Evaluate Performance And Maximize RevenueHow To Use Benchmarking To Evaluate Performance And Maximize Revenue
How To Use Benchmarking To Evaluate Performance And Maximize RevenueGoMio.com
 
Delivering Results: How Do You Report User Research Findings?
Delivering Results: How Do You Report User Research Findings?Delivering Results: How Do You Report User Research Findings?
Delivering Results: How Do You Report User Research Findings?Bob Thomas
 

Semelhante a Arguing from a Point of View (15)

Identifying consumers’ arguments in text swaie at ekaw 2012 10-09
Identifying consumers’ arguments in text swaie at ekaw 2012 10-09Identifying consumers’ arguments in text swaie at ekaw 2012 10-09
Identifying consumers’ arguments in text swaie at ekaw 2012 10-09
 
Semi-automated argumentation analysis of online product reviews--COMMA 2012-0...
Semi-automated argumentation analysis of online product reviews--COMMA 2012-0...Semi-automated argumentation analysis of online product reviews--COMMA 2012-0...
Semi-automated argumentation analysis of online product reviews--COMMA 2012-0...
 
Identifying arguments for evaluation using an argument explorer - London Argu...
Identifying arguments for evaluation using an argument explorer - London Argu...Identifying arguments for evaluation using an argument explorer - London Argu...
Identifying arguments for evaluation using an argument explorer - London Argu...
 
Kdd 2014 Tutorial - the recommender problem revisited
Kdd 2014 Tutorial -  the recommender problem revisitedKdd 2014 Tutorial -  the recommender problem revisited
Kdd 2014 Tutorial - the recommender problem revisited
 
Using trust-aware strategic agents for a self-organising computing grid
Using trust-aware strategic agents for a self-organising computing gridUsing trust-aware strategic agents for a self-organising computing grid
Using trust-aware strategic agents for a self-organising computing grid
 
Interactive XAI for ODSC East 2023
Interactive XAI for ODSC East 2023Interactive XAI for ODSC East 2023
Interactive XAI for ODSC East 2023
 
A Multidomain and Multilingual Conceptual Data Model for Online Reviews Repre...
A Multidomain and Multilingual Conceptual Data Model for Online Reviews Repre...A Multidomain and Multilingual Conceptual Data Model for Online Reviews Repre...
A Multidomain and Multilingual Conceptual Data Model for Online Reviews Repre...
 
Jumpstart: Introduction to Schema Design
Jumpstart: Introduction to Schema DesignJumpstart: Introduction to Schema Design
Jumpstart: Introduction to Schema Design
 
Opinion-based User Profile Modeling for Contextual Suggestions
Opinion-based User Profile Modeling for Contextual SuggestionsOpinion-based User Profile Modeling for Contextual Suggestions
Opinion-based User Profile Modeling for Contextual Suggestions
 
Intro to R and Data Mining 2012 09 27
Intro to R and Data Mining 2012 09 27Intro to R and Data Mining 2012 09 27
Intro to R and Data Mining 2012 09 27
 
User Generated Video Reviews by Hotel Guests
User Generated Video Reviews by Hotel GuestsUser Generated Video Reviews by Hotel Guests
User Generated Video Reviews by Hotel Guests
 
World Usability Day 2014: Engagement (Master Deck)
World Usability Day 2014: Engagement (Master Deck)World Usability Day 2014: Engagement (Master Deck)
World Usability Day 2014: Engagement (Master Deck)
 
How To Use Benchmarking To Evaluate Performance And Maximize Revenue
How To Use Benchmarking To Evaluate Performance And Maximize RevenueHow To Use Benchmarking To Evaluate Performance And Maximize Revenue
How To Use Benchmarking To Evaluate Performance And Maximize Revenue
 
Delivering Results: How Do You Report User Research Findings?
Delivering Results: How Do You Report User Research Findings?Delivering Results: How Do You Report User Research Findings?
Delivering Results: How Do You Report User Research Findings?
 
Agile/Lean/Rapid UX
Agile/Lean/Rapid UXAgile/Lean/Rapid UX
Agile/Lean/Rapid UX
 

Arguing from a Point of View

  • 1. Arguing from a Point of View Adam Wyner1 and Jodi Schneider2 1 - Department of Computer Science, University of Liverpool 2 – Digital Enterprise Research Institute, National University of Ireland First International Conference on Agreement Technologies Centre for Advanced Academic Studies, University of Zagreb Dubrovnik, Croatia October 16, 2012
  • 2. Overview • Hotel reviews are a source of arguments. • Point of view is needed to evaluate arguments such as – The hotel is in an excellent location. • Therefore we relativise evaluative statements based on point of view. • The key point: evaluative statements can be justified using instantiated argumentation schemes relative to a user and a domain model. October 16, 2012 Wyner and Schneider, AT 2012 1
  • 3. Hotel use case October 16, 2012 Wyner and Schneider, AT 2012 2
  • 4. Positive reviews October 16, 2012 Wyner and Schneider, AT 2012 3
  • 5. Negative reviews October 16, 2012 Wyner and Schneider, AT 2012 4
  • 6. TMI • How much 'bad' spoils what amount of 'good'? • How do the scores relate to the content? How does the content justify or argue for the score given? • How do the comments relate to one another? Linear text & lists of comments aren’t rich enough: Elaborate network of point and counterpoint. October 16, 2012 Wyner and Schneider, AT 2012 5
  • 7. It's all about YOU! People don't just want “information” They want information that is • relevant to them • appeals to them • sees things from their point of view. October 16, 2012 Wyner and Schneider, AT 2012 6
  • 8. Goal • To support relativised argumentation derived from distributed, inconsistent information. October 16, 2012 Wyner and Schneider, AT 2012 7
  • 9. Evaluative expressions use case - client and travel agent - I'm going to a conference in venue X in Valencia and need a hotel room. Bill Hotel Valencia is in an excellent location. Travel agent Why do you say it is an excellent location? The hotel is a kilometer from the venue X. And the hotel is in the old part of the city. OK, please book it. October 16, 2012 Wyner and Schneider, AT 2012 8
  • 10. Evaluative expressions use case - client and travel agent - I'm going to a conference in venue X in Valencia and need a hotel room. Jill Hotel Valencia is in an excellent location. Why do you say it is an excellent location? The hotel is a kilometer from the venue X. And the hotel is in the old part of the city. But it is a noisy and trashy old part. And it is too far. Please find me something else. October 16, 2012 Wyner and Schneider, AT 2012 9
  • 11. Argument evaluation is user-relative • Bill & Jill receive the same argument from the travel agent but evaluate it differently. • Given the premises – The hotel is a kilometer from the venue X. And the hotel is in the old part of the city. • Bill has accepted the claim – Hotel Valencia is in an excellent location. • Given the same premises, Jill has not accepted the claim (and doesn't even agree with all the premises). • Different ways to argue for and against the same claim. October 16, 2012 Wyner and Schneider, AT 2012 10
  • 12. Approach • Argumentation schemes are key – Normative patterns of defeasible reasoning. – Variables can be seen as targets for information extraction. Could use text analysis to instantiate. – Evaluate instantiated arguments using argumentation frameworks. • Relativise the instantiated arguments to a user. October 16, 2012 Wyner and Schneider, AT 2012 11
  • 13. Argumentation Schemes Overview • Example scheme from the literature “Credible source”: – Instantiated – Abstract – Questions used to critique the argument • Two new schemes for our use case – “Evaluation of location” – “Evaluation of quality” October 16, 2012 Wyner and Schneider, AT 2012 12
  • 14. Argumentation scheme example - instantiated - • Normative patterns of defeasible reasoning: – Dr. Rose is an expert about road safety; – Dr. Rose asserts that having more speed cameras will save more lives; – Having more speed cameras will save lives is a statement concerning road safety; – Dr. Rose is credible about road safety; – and Dr. Rose is reliable; – Therefore, it is presumably true that having more speed cameras will save more lives. October 16, 2012 Wyner and Schneider, AT 2012 13
  • 15. Argumentation scheme example - abstracted - • Normative patterns of defeasible reasoning: – X is an expert about Y; – X asserts Z; – Z is a statement concerning Y; – X is credible about Y; – and X is reliable; – therefore, it is presumably true that Z. October 16, 2012 Wyner and Schneider, AT 2012 14
  • 16. Argumentation scheme example - critique - • Questions used to critique the argument: – How credible is X as an expert source? – Is the claim about Z consistent with what other experts assert? – Is X’s assertion based on evidence? – Others.... October 16, 2012 Wyner and Schneider, AT 2012 15
  • 17. Use case elements • New argumentation schemes: – Evaluation of location. – Evaluation of quality. • Instantiate schemes relative to a user model. • Domain and evaluative terminology. • User model – selection from domain terminology plus some terminology for parameters, context of use, constraints.... October 16, 2012 Wyner and Schneider, AT 2012 16
  • 18. User Information In this paper, we represent user models by terminology and instantiated schemes. In other work, we add these components. October 16, 2012 Wyner and Schneider, AT 2012 17
  • 19. “Evaluation of location” arg. scheme - abstract - October 16, 2012 Wyner and Schneider, AT 2012 18
  • 20. “Evaluation of location” arg. scheme - Instantiating for Bill & Jill - October 16, 2012 Wyner and Schneider, AT 2012 19
  • 21. “Evaluation of quality” arg. scheme -abstract- October 16, 2012 Wyner and Schneider, AT 2012 20
  • 22. “Evaluation of quality” arg. scheme -Instantiating for Bill- October 16, 2012 Wyner and Schneider, AT 2012 21
  • 23. Instantiating for Jill October 16, 2012 Wyner and Schneider, AT 2012 22
  • 24. Use case elements • Argumentation schemes: – Evaluation of location. – Evaluation of quality. • Instantiate schemes relative to a user model. • Domain and evaluative terminology. • User model – selection from terminology and instantiated schemes. October 16, 2012 Wyner and Schneider, AT 2012 23
  • 25. Domain and evaluative terminology October 16, 2012 Wyner and Schneider, AT 2012 24
  • 26. User-associated inference • If the instantiations of both argumentation schemes are acceptable to the user, then the user has a justification to book the hotel. • For us, the model of the user can be given in terms of a logical language – the terminology and the schemes instantiated with that terminology. • Arguing about the instantiations, e.g. Jill's criticism of the travel agent's proposition, is a meta-argument about the contents of the user model. October 16, 2012 Wyner and Schneider, AT 2012 25
  • 27. Argumentation frameworks & text analysis • This paper is part of a larger work on the argumentation pipeline, from textual source to abstract argumentation. • Introduces new schemes and instantiates them relative to a user. • Other parts: – We have a text analytic tool (GATE) to support the extraction of relevant information from the source. – We have a proposal for integrating this with argumentation frameworks. October 16, 2012 Wyner and Schneider, AT 2012 26
  • 28. Argumentation pipeline Source Text Instantiated Abstract Argumentation No fresh orange juice at Argumentation Schemes Frameworks breakfast and besides terrible filter coffee extra payment for cappuchino etc... No wifi in the rooms (says so in description, but still...). AS1: .... Relate Extract text schemes to Very impressive hotel with to schemes arguments. stunning views. Staff were attentive - especially the bell boys. 5 min bus journey to the AS2: .... old town or 15 min walk. The room was very comfortable. If u want to stay with comfort I would never recommend this hotel on arrival I was waiting AS3: .... my room from 14.00 till 16.00, but again they gave me a room with two separate beds ignoring my comments in the booking (one king bed and big bathtube) October 16, 2012 Wyner and Schneider, AT 2012 27
  • 29. Consumer argumentation scheme Variables in schemes as targets for extraction. Premises: • Camera X has property P. • Property P promotes value V for agent A. Conclusion: • Agent A should Action Camera X. October 16, 2012 Wyner and Schneider, AT 2012 28
  • 30. Identifying and extracting text • Annotate text: – Simple or complex annotations. – Highlight annotations with – Search for and extract text by annotation. • GATE “General Architecture for Text Engineering”. – Works with large corpora of text. – Rule-based or machine-learning approaches. October 16, 2012 Wyner and Schneider, AT 2012 29
  • 31. , , October 16, 2012 Wyner and Schneider, AT 2012 30
  • 32. Query for patterns October 16, 2012 Wyner and Schneider, AT 2012 31
  • 33. An argument for buying the camera Premises: The pictures are perfectly exposed. The pictures are well-focused. No camera shake. Good video quality. Each of these properties promotes image quality. Conclusion: (You, the reader,) should buy the CanonSX220. October 16, 2012 Wyner and Schneider, AT 2012 32
  • 34. An argument for NOT buying the camera Premises: The colour is poor when using the flash. The images are not crisp when using the flash. The flash causes a shadow. Each of these properties demotes image quality. Conclusion: (You, the reader,) should not buy the CanonSX220. October 16, 2012 Wyner and Schneider, AT 2012 33
  • 35. Counterarguments to the premises of “Don’t buy” The colour is poor when using the flash. For good colour, use the colour setting, not the flash. The images are not crisp when using the flash. No need to use flash even in low light. The flash causes a shadow. There is a corrective video about the flash shadow. October 16, 2012 Wyner and Schneider, AT 2012 34
  • 36. Argumentation Frameworks • <Arguments, Relation>, where arguments are atomic nodes and the relation is attack. • Calculate the sets of nodes that are 'compatible'. • Articulate nodes with a logical language of literals and rules, where attack is contrariness between Preferred Extension of the AF. expressions of the language. October 16, 2012 Wyner and Schneider, AT 2012 35
  • 37. Future work • User model formalisation and meta-argumentation. • Text analysis for this set of data. • Tool refinement. • Add ontology modules to the tool. • Multi-critierial argumentation – properties ascribed to the argument vs. premises of the argument. October 16, 2012 Wyner and Schneider, AT 2012 36
  • 38. Related Papers • Wyner, van Engers, and Hunter (2010). "Working on the Argument Pipeline: Through Flow Issues between Natural Language Argument, Instantiated Arguments, and Argumentation Frameworks", Workshop on Computational Models of Natural Argument (CMNA). • Wyner, Schneider, Atkinson, and Bench-Capon (2012). ''Semi- automated argumentation analysis of online product reviews'', Conference on Computational Models of Argument (COMMA). • Schneider and Wyner (2012). ''Identifying consumers' arguments in text'', Workshop on Semantic Web and Information Extraction (SWAIE at EKAW). • Schneider, Davis, and Wyner (2012). ''Dimensions of argumentation in social media'', Conference on Knowledge Engineering and Knowledge Management (EKAW). October 16, 2012 Wyner and Schneider, AT 2012 37
  • 39. Acknowledgements • FP7-ICT-2009-4 Programme, IMPACT Project, Grant Agreement Number 247228. • Science Foundation Ireland Grant No. SFI/08/CE/I1380 (Líon- 2). Short-term Scientific Mission grant from COST Action IC0801 on Agreement Technologies. SFI Short Term Travel Fellowship. October 16, 2012 Wyner and Schneider, AT 2012 38
  • 40. Thanks for your attention! • Questions? • Contacts: – Adam Wyner adam@wyner.info – Jodi Schneider jschneider@pobox.com October 16, 2012 Wyner and Schneider, AT 2012 39

Notas do Editor

  1. Tuesday, October 16, 2012
  2. This doesn’t mentionextracting arguments from source text
  3. For or against the hotel? Or the hotel RATING?
  4. Not sure if “Different ways to argue for and against the same claim.” is what we want…Isn’t it really that there are Different ways to EVALUATE ARGUMENTS for and against the same claim?
  5. Any hotel domain example that would be easy here?Why not just spend more time on one of our examples? I think that would make sense….
  6. Might focus just on this paper here
  7. Not sure why you call this co-variance
  8. Haven’t shown the pipeline up till now – is “middle” clear enough? Maybe this is what you say when you’re showing the pipeline? Not sure if this slide is essential then.
  9. Pipeline picture – linear text, schemes, argument reconstruction, abstract evaluation.
  10. Colors represent annotations in the text. We can then search for a large body of text
  11. Leave camera implicit in the examples for brevity.
  12. We have an argument for buying the camera, an argument for not buying the camera. They rebut each other.We have attacks on the premises for “don’t buy the camera”. The argument for not buying the camera is defeated; the argument for buying the camera stands. So you should buy the camera.
  13. Any future work specific to this?