hybrid models for text and graphs
DESCRIPTION
Hybrid Models for Text and Graphs. 10/23/2012 Analysis of Social Media. Formal Primary purpose: Inform “ typical reader ” about recent events Broad audience: Explicitly establish shared context with reader Ambiguity often avoided. Informal Many purposes: Entertain, connect, persuade… - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/1.jpg)
Hybrid Models for Text and Graphs
10/23/2012Analysis of Social Media
![Page 2: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/2.jpg)
Newswire Text
• Formal• Primary purpose:
– Inform “typical reader” about recent events
• Broad audience:– Explicitly establish
shared context with reader
– Ambiguity often avoided
• Informal• Many purposes:
– Entertain, connect, persuade…
• Narrow audience:– Friends and
colleagues– Shared context
already established– Many statements are
ambiguous out of social context
Social Media Text
![Page 3: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/3.jpg)
Newswire Text
• Goals of analysis:– Extract information
about events from text
– “Understanding” text requires understanding “typical reader”
• conventions for communicating with him/her
• Prior knowledge, background, …
• Goals of analysis:– Very diverse– Evaluation is difficult
• And requires revisiting often as goals evolve
– Often “understanding” social text requires understanding a community
Social Media Text
![Page 4: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/4.jpg)
Outline
• Tools for analysis of text– Probabilistic models for text,
communities, and time• Mixture models and LDA models for text• LDA extensions to model hyperlink structure• LDA extensions to model time
![Page 5: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/5.jpg)
Introduction to Topic Models
• Mixture model: unsupervised naïve Bayes model
C
W
NM
• Joint probability of words and classes:
• But classes are not visible:Z
![Page 6: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/6.jpg)
Introduction to Topic Models
JMLR, 2003
![Page 7: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/7.jpg)
Introduction to Topic Models
• Latent Dirichlet Allocation
z
w
M
N
• For each document d = 1,,M
• Generate d ~ Dir(.| )
• For each position n = 1,, Nd
• generate zn ~ Mult( . | d)
• generate wn ~ Mult( .| zn)
![Page 8: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/8.jpg)
Introduction to Topic Models
• Latent Dirichlet Allocation– Overcomes some technical issues with PLSA
• PLSA only estimates mixing parameters for training docs
– Parameter learning is more complicated:• Gibbs Sampling: easy to program, often slow• Variational EM
![Page 9: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/9.jpg)
Introduction to Topic Models
• Perplexity comparison of various models
Unigram
Mixture model
PLSA
LDALower is better
![Page 10: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/10.jpg)
Introduction to Topic Models• Prediction accuracy for classification using
learning with topic-models as features
Higher is better
![Page 11: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/11.jpg)
Before LDA….LSA and pLSA
![Page 12: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/12.jpg)
Introduction to Topic Models
• Probabilistic Latent Semantic Analysis Model
d
z
w
M
• Select document d ~ Mult()
• For each position n = 1,, Nd
• generate zn ~ Mult( _ | d)
• generate wn ~ Mult( _ | zn)
d
N
Topic distribution
PLSA model:
• each word is generated by a single unknown multinomial distribution of words, each document is mixed by d
• need to estimate d for each d overfitting is easy
LDA:
•integrate out d and only estimate
![Page 13: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/13.jpg)
Introduction to Topic Models
• PLSA topics (TDT-1 corpus)
![Page 14: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/14.jpg)
Outline
• Tools for analysis of text– Probabilistic models for text, communities,
and time• Mixture models and LDA models for text•LDA extensions to model hyperlink
structure• LDA extensions to model time
– Alternative framework based on graph analysis to model time & community• Preliminary results & tradeoffs
• Discussion of results & challenges
![Page 15: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/15.jpg)
Hyperlink modeling using PLSA
![Page 16: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/16.jpg)
Hyperlink modeling using PLSA
[Cohn and Hoffman, NIPS, 2001]
d
z
w
M
d
N
z
c
• Select document d ~ Mult()
• For each position n = 1,, Nd
• generate zn ~ Mult( . | d)
• generate wn ~ Mult( . | zn)
• For each citation j = 1,, Ld
• generate zj ~ Mult( . | d)
• generate cj ~ Mult( . | zj)L
![Page 17: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/17.jpg)
Hyperlink modeling using PLSA[Cohn and Hoffman, NIPS, 2001]
d
z
w
M
d
N
z
c
L
PLSA likelihood:
New likelihood:
Learning using EM
![Page 18: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/18.jpg)
Hyperlink modeling using PLSA[Cohn and Hoffman, NIPS, 2001]
Heuristic:
0 · · 1 determines the relative importance of content and hyperlinks
(1-)
![Page 19: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/19.jpg)
Hyperlink modeling using PLSA[Cohn and Hoffman, NIPS, 2001]
• Experiments: Text Classification• Datasets:
– Web KB• 6000 CS dept web pages with hyperlinks• 6 Classes: faculty, course, student, staff, etc.
– Cora• 2000 Machine learning abstracts with citations• 7 classes: sub-areas of machine learning
• Methodology:– Learn the model on complete data and obtain d for each
document– Test documents classified into the label of the nearest
neighbor in training set– Distance measured as cosine similarity in the space– Measure the performance as a function of
![Page 20: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/20.jpg)
Hyperlink modeling using PLSA[Cohn and Hoffman, NIPS, 2001]
• Classification performance
Hyperlink content link content
![Page 21: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/21.jpg)
Hyperlink modeling using LDA
![Page 22: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/22.jpg)
Hyperlink modeling using LinkLDA[Erosheva, Fienberg, Lafferty, PNAS, 2004]
z
w
M
N
• For each document d = 1,,M
• Generate d ~ Dir(¢ | )
• For each position n = 1,, Nd
• generate zn ~ Mult( . | d)
• generate wn ~ Mult( . | zn)
•For each citation j = 1,, Ld
• generate zj ~ Mult( . | d)
• generate cj ~ Mult( . | zj)
z
c
L
Learning using variational EM
![Page 23: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/23.jpg)
Hyperlink modeling using LDA[Erosheva, Fienberg, Lafferty, PNAS, 2004]
![Page 24: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/24.jpg)
Newswire Text
• Goals of analysis:– Extract information
about events from text
– “Understanding” text requires understanding “typical reader”
• conventions for communicating with him/her
• Prior knowledge, background, …
• Goals of analysis:– Very diverse– Evaluation is
difficult• And requires revisiting
often as goals evolve
– Often “understanding” social text requires understanding a community
Social Media Text
Science as a testbed for social text: an open community which we understand
![Page 25: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/25.jpg)
Author-Topic Model for Scientific Literature
![Page 26: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/26.jpg)
Author-Topic Model for Scientific Literature[Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004]
z
w
M
N
• For each author a = 1,,A
• Generate a ~ Dir(. | )
• For each topic k = 1,,K
• Generate k ~ Dir( . | )
•For each document d = 1,,M
• For each position n = 1,, Nd
•Generate author x ~ Unif(. | ad)
• generate zn ~ Mult(. | a)
• generate wn ~ Mult(. | zn)
x
a
A
P
K
![Page 27: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/27.jpg)
Author-Topic Model for Scientific Literature [Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004]
• Perplexity results
![Page 28: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/28.jpg)
Author-Topic Model for Scientific Literature [Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004]• Topic-Author visualization
![Page 29: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/29.jpg)
Author-Topic Model for Scientific Literature[Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004]
• Application 1: Author similarity
![Page 30: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/30.jpg)
Author-Topic Model for Scientific Literature [Rozen-Zvi, Griffiths, Steyvers, Smyth UAI, 2004]
• Application 2: Author entropy
![Page 31: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/31.jpg)
Labeled LDA: [Ramage, Hall, Nallapati, Manning, EMNLP 2009]
![Page 32: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/32.jpg)
Labeled LDADel.icio.us tags as labels for documents
![Page 33: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/33.jpg)
Labeled LDA
![Page 34: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/34.jpg)
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05]
![Page 35: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/35.jpg)
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05]
Gibbs sampling
![Page 36: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/36.jpg)
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05]
• Datasets– Enron email data
• 23,488 messages between 147 users
– McCallum’s personal email• 23,488(?) messages with 128 authors
![Page 37: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/37.jpg)
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05]
• Topic Visualization: Enron set
![Page 38: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/38.jpg)
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05]
• Topic Visualization: McCallum’s data
![Page 39: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/39.jpg)
Author-Topic-Recipient model for email data [McCallum, Corrada-Emmanuel,Wang, ICJAI’05]
![Page 40: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/40.jpg)
Models of hypertext for blogs [ICWSM 2008]
Ramesh Nallapati
me
Amr Ahmed Eric Xing
![Page 41: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/41.jpg)
LinkLDA model for citing documentsVariant of PLSA model for cited documentsTopics are shared between citing, citedLinks depend on topics in two documents
Link-PLSA-LDA
![Page 42: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/42.jpg)
Experiments
• 8.4M blog postings in Nielsen/Buzzmetrics corpus– Collected over three weeks summer 2005
• Selected all postings with >=2 inlinks or >=2 outlinks– 2248 citing (2+ outlinks), 1777 cited documents (2+
inlinks)– Only 68 in both sets, which are duplicated
• Fit model using variational EM
![Page 43: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/43.jpg)
Topics in blogs
Model can answer questions like: which blogs are most likely to be cited when discussing topic z?
![Page 44: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/44.jpg)
Topics in blogs
Model can be evaluated by predicting which links an author will include in a an article
Lower is better
Link-PLSA-LDA
Link-LDA
![Page 45: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/45.jpg)
Another model: Pairwise Link-LDA
z
w
N
z
w
N
z z
c
• LDA for both cited and citing documents • Generate an indicator for every pair of docs
• Vs. generating pairs of docs
•Link depends on the mixing components (’s)
• stochastic block model
![Page 46: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/46.jpg)
Pairwise Link-LDA supports new inferences…
…but doesn’t perform better on link prediction
![Page 47: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/47.jpg)
Outline
• Tools for analysis of text– Probabilistic models for text,
communities, and time• Mixture models and LDA models for text•LDA extensions to model hyperlink
structure– Observation: these models can be used for
many purposes…
• LDA extensions to model time
– Alternative framework based on graph analysis to model time & community
• Discussion of results & challenges
![Page 48: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/48.jpg)
![Page 49: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/49.jpg)
Authors are using a number of clever tricks for inference….
![Page 50: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/50.jpg)
![Page 51: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/51.jpg)
![Page 52: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/52.jpg)
![Page 53: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/53.jpg)
Predicting Response to Political Blog Posts with Topic Models [NAACL ’09]
Tae Yano Noah Smith
![Page 54: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/54.jpg)
54
Political blogs and and comments
Comment style is casual, creative,less carefully edited
Posts are often coupled with comment sections
![Page 55: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/55.jpg)
Political blogs and comments
• Most of the text associated with large “A-list” community blogs is comments– 5-20x as many words in comments as in text for
the 5 sites considered in Yano et al.
• A large part of socially-created commentary in the blogosphere is comments.– Not blog blog hyperlinks
• Comments do not just echo the post
![Page 56: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/56.jpg)
Modeling political blogs
Our political blog model:
CommentLDA
D = # of documents; N = # of words in post; M = # of words in comments
z, z` = topicw = word (in post)w`= word (in comments)u = user
![Page 57: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/57.jpg)
Modeling political blogs
Our proposed political blog model:
CommentLDALHS is vanilla LDA
D = # of documents; N = # of words in post; M = # of words in comments
![Page 58: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/58.jpg)
Modeling political blogs
Our proposed political blog model:
CommentLDA
RHS to capture the generation of reaction separately from the post body
Two separate sets of word distributions
D = # of documents; N = # of words in post; M = # of words in comments
Two chambers share the same topic-mixture
![Page 59: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/59.jpg)
Modeling political blogs
Our proposed political blog model:
CommentLDA
User IDs of the commenters as a part of comment text
generate the wordsin the comment section
D = # of documents; N = # of words in post; M = # of words in comments
![Page 60: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/60.jpg)
Modeling political blogs
Another model we tried:
CommentLDA
This is a model agnostic to the words in the comment section!
D = # of documents; N = # of words in post; M = # of words in comments
Took out the words from the comment section!
The model is structurally equivalent to the LinkLDA from (Erosheva et al., 2004)
![Page 61: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/61.jpg)
61
Topic discovery - Matthew Yglesias (MY) site
![Page 62: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/62.jpg)
62
Topic discovery - Matthew Yglesias (MY) site
![Page 63: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/63.jpg)
63
Topic discovery - Matthew Yglesias (MY) site
![Page 64: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/64.jpg)
64
• LinkLDA and CommentLDA consistently outperform baseline models• Neither consistently outperforms the other.
Comment prediction
20.54 %
16.92 % 32.06
%
Comment LDA (R)
Link LDA (R) Link LDA (C)
user prediction: Precision at top 10From left to right: Link LDA(-v, -r,-c) Cmnt LDA (-v, -r, -c), Baseline (Freq, NB)
(CB)
(MY)
(RS)
![Page 65: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/65.jpg)
Document modeling with Latent Dirichlet Allocation (LDA)
z
w
M
N
• For each document d = 1,,M
• Generate d ~ Dir(. | )
• For each position n = 1,, Nd
• generate zn ~ Mult( . | d)
• generate wn ~ Mult( . | zn)
![Page 66: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/66.jpg)
Modeling Citation Influences
![Page 67: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/67.jpg)
Modeling Citation Influences[Dietz, Bickel, Scheffer, ICML 2007]
• Copycat model of citation influence
c is a cited document s is a coin toss to mix γ and
plaigarism
innovation
![Page 68: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/68.jpg)
s is a coin toss to mix γ and
![Page 69: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/69.jpg)
Modeling Citation Influences[Dietz, Bickel, Scheffer, ICML 2007]
• Citation influence graph for LDA paper
![Page 70: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/70.jpg)
Modeling Citation Influences
![Page 71: Hybrid Models for Text and Graphs](https://reader035.vdocuments.site/reader035/viewer/2022062321/56812dd4550346895d931e09/html5/thumbnails/71.jpg)
Modeling Citation Influences
User study: self-reported citation influence on Likert scale
LDA-post is Prob(cited doc|paper)
LDA-js is Jensen-Shannon dist in topic space