Literature Review

advertisement
Carlos Figueiredo
Emotional Response and
Bridging ties on Social Networks

Motivations

Overview

Research Question
intervene in the Surprise Response

Literature Review
in Information Sharing

Research Methods

Desired Contributions

Suggestions
Interpreting how Individual Attributes
and Social Graph Properties
Digital Media Research Methods
Porto, DiMe 24th May 2013
Motivation
 Make a contribution to solve the problem of
the ‘Portfolio Effect’ on Recommender System
(e.g. content recommendation).
 Measure novelty in an information flow.
 Propose a methodology to measure and
predict the delivering of novelty.
Overview
Recommender Systems

Content-Based
“You bought, liked or
shared a bunch of
things like Y”
(1)

Collaborative
Filtering
“People like you
bought, liked or
shared Y”

Trust-Based CF
“People who bought,
liked or shared X
also bought, liked or
shared Y”
(2)
e.g. Amazon
(1) coding the genome of each song;
(2) listen to other users’ radio stations
(Friends, Neighbors, Groups)
e.g. Friend Network (Fb);
Reputation Network (Ebay)
Overview
Recommenders
my proposal
 Recommendation based
on novelty.
 Novelty delivering based
on users attributes and
in the social graph
properties.
 Surprise response as a
proxy of the Novelty to
design the framework.
Literature Review
topics of research
 Novelty delivering.

Bridging ties
 Dimensions implied on novelty.

Emotional response (Surprise)

Tie strength (between receptor and source of information)

Homophily (similarities)

Centrality (structural location in the social network)
Literature Review
novelty delivering
Bridging Ties
 The Strength of Weak
Ties (Granovetter 1973)
 All bridges are weak ties.
Not all weak ties are
bridges.
 Strong ties can be a bridge
if such ties do not share
ties with other individuals
of the same clique.
How to deliver novelty?
Literature Review
novelty delivering
Bridging Ties
 Structural Holes
(Burt 1992)
 Tie strength does not determine
the information potential
brought by a bridging tie.
 Structural holes linked by actors
determine the information
potential.
 Ties established by nonredundant links between actors.
 Weak ties and Structural holes
Structural Holes and Weak Ties
(Burt 1992)
(McEvily et al. 1999)
Literature Review
dimensions implied on novelty
Emotional response
Can I predict novelty?
 Ten primordial emotions (DES scale of Izard)
(Izard 1991)
 Emotions start with a process of cognitive
appraisal (Finkenauer et al. 1998).
 Novelty is one fundamental type of appraisal
(e.g., Scherer 1984; Smith and Ellsworth 1987).
 Surprise is a specific consequence of the
appraisal of novelty (Finkenauer et al. 1998).
Literature Review
dimensions implied on novelty
Measuring tie strength
What makes a tie be strong?
 Amount of time, intimacy, intensity and
reciprocal services (Granovetter 1973).
 Closeness
(Marsden et al. 1984).
 Emotional support
(Wellman and Wortley 1990).
 Intensity and valence
(Petrosky 2011).
Literature Review
dimensions implied on novelty
Homophily
Why some weak ties are not bridges?
 Demographic and Attitudinal similarities
 Status homophily
 Background factors (economic status), gender, religion
and ethnicity (McPherson et al. 2001).
 Attitudinal homophily
 Perceived Homophily Measure (McCroskey et al. 1975;
2006).
Literature Review
dimensions implied on novelty
Homophily
 Cognitive similarities based on:
 Music genres
Reflective and Complex, Intense and Rebellious, Upbeat
and Conventional, and Energetic and Rhythmic (Rentfrow
and Gosling 2003).
 Emotional reaction to music genres.
 Similarities on information interests. Similarities on
emotional reaction to the information accessed.
Equal preferences on content posted (survey).
Literature Review
dimensions implied on novelty
Centrality
Can I predict structural location?
 Degree; betweeness; closeness
(Freeman 1979)
 Centrality and information flow
(Borgatti et al. 2009;
Mori et al. 2010)
 Centrality and friendship network
Afuah 2013)
(Opshal 2010;
Research questions
 Does surprise define bridging ties?
 Can similarities in individual attributes and
structural location in the network determine the
stimulus of surprise?
Hypothesis
 H1 – Is more likely that an individual become
surprised by a weak tie than by a strong tie.
 H2 – The non redundancy between ties is
predictor of surprise.
 H3 – The degree of homophily justifies the
conditions that make someone weakly tied to
another person acting as a bridge.
 H4 – Users' popularity is independent of the
novelty appraisal by the receptors.
Research Methods
 Survey
 Emotional response, particularly on surprise
 Ties strength
 Homophily
 SPSS
 Statistics (Logistic regression; Multivariate regression;
Mann-Whitney test)
Research Methods
 NodeXl
 Centrality variables (degree; betweeness; closeness)
Graph 1 - Emotional reaction to
information sharing
Graph 2 – Friendship network
Desired contribution
 Emotional reaction as a relevant property for
social networks analysis.
 A methodology that identifies bridging ties and
characterizes the delivering of novelty based in
the stimulus of surprise.

 A framework for recommendation systems
based on weak ties as recommenders of
surprising information or of new perspectives.
Suggestions
 Comments are welcome…
 Thank you.
Carlos Figueiredo
Download