Argumentation – On the Social Web – Problem Space

View Github Repository Open presentation in a new window

Landric

See all presentation from Landric

Argumentation – On the Social Web – Problem Space

0 1

argumentation-presentation

Short presentation on argumentation and the social web

On Github Landric / argumentation-presentation

Argumentation

On the Social Web

Tom Blount

Slides available at: http://tomblount.co.uk/argumentation-presentation

Introduction

  • Problem space
  • Argumentation
  • Social web
  • Initial findings
  • Future work
In this presentation, I'll go over the problem space, explain some of the work that has been done in the areas of argumentation and the social web, and cover my initial findings and future work.

Problem Space

Have you ever argued online?

Did you "win"?

How did it make you feel?

How many of you have ever gotten involved in an argument on the internet? How did it make you feel? Pleased? Or frustrated?

Problem Space

  • Cyberbullying
  • Perception of the web
  • "Wicked" problems
Cyberbullying is becoming an increasing problem, which in turn is colouring a negative perception of the web I don't think the web deserves this, the web is a brilliant technology: in fact, if we can improve argument quality, we could crowd source solutions to so-called "wicked" problems; for example, the issue of climate change, or political referendums

Argumentation

Argumentation is a key element of human communication

Image via Wikimedia So that's not to say all argumentation is a bad thing - resolving a difference of opinions is a necessary part of human communication Argumentation has been studied since the dawn of civilisation - Aristotle and Plato (who you can see here) discussed it at length In particular, there are lots of formal models of argumentation

Formal Models of Argumentation

  • Predicate/propositional logic
  • Set theory
  • Graph theory
  • Legal proceedings
  • Academic proofs
  • Cognitive ergonomics
People have tried to model argument using a number of formal techniques - logic, graph theory and so on These can be used to visualise the trail of evidence in court, the reasoning in an academic paper or to elicit knowledge from a collection of experts

The Toulmin Model

Image adapted from Toulmin (1958) This is the Toulmin model - a diagram representing the structure of an argument A Claim is the conclusion someone is hoping to draw - for example, "I am a British citizen" A Qualifier shows the certainty of an argument - "definitely", "probably", "quite likely", etc. Data are the facts they use to draw this Claim - "I hold a British passport" Warrant is the implicit joining of these two statements - "The holder of a British passport will be a British citizen" Backing is further proof of the Warrant A rebuttal is a foreseen counter-argument - "Unless I am a spy of a foreign power"

The Toulmin Model

A Claim is the conclusion someone is hoping to draw - for example, "I am a British citizen"

A Qualifier shows the certainty of an argument - "definitely", "probably", "quite likely", etc.

Data are the facts they use to draw this Claim - "I hold a British passport"

The Toulmin Model

The Warrant is the implicit joining of these two statements - "The holder of a British passport will be a British citizen"

Backing is further proof of the Warrant

A Rebuttal is a foreseen counter-argument - "Unless I am a spy of a foreign power"

The Social Web

Image via Brian Solis - (Other sizes) However - have you ever thought these things through if you've been involved in an argument online? Most people tend not to. Lets take a look at the Social Web The social web is a vast network of people and includes everything from news discussions to content creations

Social Web

  • Vast numbers of participants
  • Long lengths of time between responses
  • Ability to directly link data/evidence/citations
  • Rating systems (Facebook "Likes" or reddit "karma")
  • Restrictions on posting (144 character limit, language filters, etc.)
There are lots of features on the social web that aren't accounted for in traditional models of argument such as scale, reputation systems and so on Does this skew our perception of argumentation online?

Initial Findings

  • Simple comment scraper
  • Makes use of sentiment analysis web-service
  • Analyses multiple social web services
To get an idea of how people currently argue on the web, I wrote a very simple Python script than analyses the sentiment of comments on various websites (whether they are positive or negative)

Initial Findings

Results: Twitter These show comments on Barack Obama's twitter feed - as you can see, there are some positive areas, some negative. Roughly, there seem to be equal numbers of people agreeing and disagreeing

Initial Findings

Results: Daily Mail Here however, these comments on a Daily Mail article are overwhelmingly negative

Initial Findings

Results: 4chan These are comments from 4chan - again, mostly negative. The empty space is likely due to the fact that 4chan is an image board, so some comments won't have text to analyse

Future Work

Apply current models, to see where their deficiencies lie Create a formal model of social argumentation that captures aspects missing from previous models Use this model to determine how to improve behaviour and encourage dialectic argument on the web The first task is to analyse the current models, like Toulmin's, to see which aspects of the social web they struggle to capture (but also which areas they excel at Next is to build on these models, or develop a new model, that captures all the necessary areas Finally this model can then be used to determine how to encourage people in a more productive fashion

Summary

There are lots of models of argumentation

Most of these are not applicable to the social web

If we capture aspects of argument specific to the social web, we can improve the quality of discourse for everyone

There are lots of models of argumentation, but most of these are not applicable to the social web. If we work on modelling aspects of argument specific to the social web, we can improve the quality of discourse for everyone.

Summary

Perhaps we'll see less of this in future

Image via xkcd So perhaps, in the future, we'll see less of this

References & Further Reading

The Uses of Argument, S. E. Toulmin, (1958), University Press, Cambridge. The Art of Always Being Right: Thirty Eight Ways to Win When You Are Defeated, A. Schopenhauer and A. C. Grayling (2004), Gibson Square Books, ISBN 1-903933-61-7 A review of argumentation for the social semantic web, J. Schneider, T. Groza and A. Passant (2013). Semantic Web, 4(2) p. 159-218 And, if you're interested in the topic, these are some extra sources of information