Threshold Models of Collective Behavior1
Mark Granovetter
State University of N ew York at Stony Brook
Models of collective behavior are developed for situations where
actors have two alternatives and the costs and/0r beneﬁts of each
depend on how many ot
Information Retrieval Models:
Probabilistic Models
ChengXiang Zhai
Department of Computer Science
University of Illinois, Urbana-Champaign
Modeling Relevance:
Raodmap for Retrieval Models
Relevance constraints
Relevance
[Fang et al. 04]
Div. from Randomne
Information Retrieval Models:
Language Models
ChengXiang Zhai
Department of Computer Science
University of Illinois, Urbana-Champaign
Modeling Relevance:
Raodmap for Retrieval Models
Relevance constraints
Relevance
[Fang et al. 04]
Div. from Randomne
(Ama
Essential Probability & Statistics
(Lecture for CS598CXZ Advanced Topics in
Information Retrieval )
ChengXiang Zhai
Department of Computer Science
University of Illinois, Urbana-Champaign
1
Prob/Statistics & Text Management
Probability & statistics provi
ResearchProblems&Topics
(WebDomain)
(CS598CXZAdvancedTopicsinIRPresentation)
Jan.25,2005
ChengXiangZhai
Department of Computer Science
UniversityofIllinois,UrbanaChampaign
FacultyHomepage
Classification/Finding
Theproblemis,toclassifythefacultyhomepages
f
CS598JHM: Advanced NLP (Spring 10)
Lecture 4:
Naive Bayes
(the Frequentist approach
and the Bayesian approach)
Todays class
The task: text classication (sentiment analysis)
Assign (sentiment) label Li ! cfw_ +,! to a document Wi=(wi1.wiN).
W1= This is an
CS598JHM: Advanced NLP (Spring 10)
Parameter estimation
Given data D=HTTHTT, what is the probability ! of heads?
Lecture 3
- Maximum likelihood estimation (MLE):
Use the ! which has the highest likelihood P(D| !).
M LE = arg max P (D|)
- Maximum a posteri
CS598JHM: Advanced NLP (Spring 10)
Why should you take this class?
Many recent NLP papers use Bayesian methods.
Lecture 1:
Introduction
If you want to do research in NLP, you need to understand these papers!
You may wonder:
- What are Bayesian methods?
(o
CS598JHM: Advanced NLP (Spring 10)
Naive Bayes for text classication
The task:
Lecture 5:
More on Gibbs sampling
Assign (sentiment) label Li ! cfw_ +,! to a document Wi.
The model:
- Li = argmax L P( L | Wi ) = argmax L P( Wi | L )P( L)
- P( Wi | L ) is a
CS598JHM: Advanced NLP (Spring 10)
The binomial distribution
If p is the probability of heads, the probability of getting
exactly k heads in n independent yes/no trials is given by
the binomial distribution Bin(n,p):
Lecture 2:
Conjugate priors
n k
P (k
CS598JHM: Advanced NLP (Spring 10)
Sampling methods
Task: Compute the expectation f(x) relative to P(x)
Sampling
(Koller/Friedman 09, Ch.12)
Approximate this through sampling:
Draw a nite number of samples from P(x)
Also known as particle-based approximat
Basic Concepts in Information Theory
ChengXiang Zhai
Department of Computer Science
University of Illinois, Urbana-Champaign
1
Background on Information Theory
Developed by Claude Shannon in the 1940s
Maximizing the amount of information that can
be tra
Probabilistic Topic Models
ChengXiang Zhai
Department of Computer Science
Graduate School of Library & Information Science
Institute for Genomic Biology
Department of Statistics
University of Illinois, Urbana-Champaign
http:/www.cs.illinois.edu/homes/czha
REPORTS
here was funded by donations from anonymous private
individuals having no connection to it. This is Paleobiology
Database publication 117.
Supporting Online Material
www.sciencemag.org/cgi/content/full/329/5996/1191/DC1
Materials and Methods
The S
Modeling Information Diffusion in Implicit Networks
Jaewon Yang
EE Department, Stanford University
[email protected]
AbstractSocial media forms a central domain for the
production and dissemination of real-time information. Even
though such ows of infor
WWW 2011 Session: Diffusion
March 28April 1, 2011, Hyderabad, India
Differences in the Mechanics of Information Diffusion
Across Topics: Idioms, Political Hashtags, and Complex
Contagion on Twitter
Daniel M. Romero
Cornell University
Ithaca, NY
[email protected]
A simple model of global cascades on
random networks
Duncan J. Watts*
Department of Sociology, Columbia University New York, NY 10027
Communicated by Murray Gell-Mann, Santa Fe Institute, Santa Fe, NM, February 14, 2002 (received for review May 29, 2001)
PRL 109, 068702 (2012)
week ending
10 AUGUST 2012
PHYSICAL REVIEW LETTERS
Locating the Source of Diffusion in Large-Scale Networks
Pedro C. Pinto, Patrick Thiran, and Martin Vetterli
Ecole Polytechnique Federale de Lausanne (EPFL), Lausanne CH-1015, Swi
The Structure of Online Diffusion Networks
SHARAD GOEL, Yahoo! Research
DUNCAN J. WATTS, Yahoo! Research
DANIEL G. GOLDSTEIN, Yahoo! Research
Models of networked diffusion that are motivated by analogy with the spread of infectious disease have
been appli
2012 IEEE 12th International Conference on Data Mining
Clash of the Contagions: Cooperation and Competition in Information Diffusion
Jure Leskovec
Stanford University
[email protected]
Seth A. Myers
Stanford University
[email protected]
AbstractIn n
CatchSync : Catching Synchronized Behavior in Large
Directed Graphs
Meng Jiang123 , Peng Cui123 , Alex Beutel4 , Christos Faloutsos4 , Shiqiang Yang123
1
2
Tsinghua National Laboratory for Information Science and Technology
Department of Computer Science
WWW 2012 Session: Information Diffusion in Social Networks
April 1620, 2012, Lyon, France
The Role of Social Networks in Information Diffusion
Eytan Bakshy
Itamar Rosenn
Facebook
1601 Willow Rd.
Menlo Park, CA 94025
Facebook
1601 Willow Rd.
Menlo Park, CA
Cascading Behavior in Large Blog Graphs
Patterns and a model
Jure Leskovec, Mary McGlohon, Christos Faloutsos
Abstract
How do blogs cite and inuence each other? How do
such links evolve? Does the popularity of old blog posts
drop exponentially with time?
The Role of Compatibility in the Diffusion of Technologies
Through Social Networks
Nicole Immorlica
Jon Kleinberg
Mohammad Mahdian
Microsoft Research
Redmond WA
Dept. of Computer Science
Cornell University, Ithaca NY
Yahoo! Research
Santa Clara CA
nickle@