Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach...
Transcript of Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach...
![Page 1: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/1.jpg)
Machine Learning for Complex Social Processes
Hanna WallachUniversity of Massachusetts Amherst
![Page 3: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/3.jpg)
Hanna Wallach :: UMass Amherst :: 4
National Institutes of Health
![Page 4: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/4.jpg)
Hanna Wallach :: UMass Amherst :: 5
United States Patent System
![Page 5: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/5.jpg)
Hanna Wallach :: UMass Amherst :: 6
Representatives and Constituents
![Page 6: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/6.jpg)
Hanna Wallach :: UMass Amherst :: 7
Social Processes: Structure
![Page 7: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/7.jpg)
Hanna Wallach :: UMass Amherst :: 8
Social Processes: Content
$$$
$$$
![Page 8: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/8.jpg)
Hanna Wallach :: UMass Amherst :: 9
Social Processes: Dynamics
$$$
$$$
![Page 9: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/9.jpg)
Hanna Wallach :: UMass Amherst :: 10
Social Processes: Dynamics
$$$
$$$
![Page 10: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/10.jpg)
Hanna Wallach :: UMass Amherst :: 11
Social Processes: Dynamics
$$$
$$$
![Page 11: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/11.jpg)
Hanna Wallach :: UMass Amherst :: 12
Social Processes: Dynamics
$$$
$$$
![Page 12: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/12.jpg)
Hanna Wallach :: UMass Amherst :: 13
Social Processes: Dynamics
$$$
$$$
![Page 13: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/13.jpg)
Hanna Wallach :: UMass Amherst :: 14
Modeling Social Processes
“Policy-makers or computer scientists may be interested in finding the needle in the haystack (such as a potential terrorist threat or the right web page to display from a search), but social scientists are more commonly interested in characterizing the haystack.”
― King & Hopkins, 2010
![Page 14: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/14.jpg)
Hanna Wallach :: UMass Amherst :: 15
Predictive Analyses
$$$ ???
![Page 15: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/15.jpg)
Hanna Wallach :: UMass Amherst :: 16
Explanatory Analyses
$$$???
![Page 16: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/16.jpg)
Hanna Wallach :: UMass Amherst :: 17
Exploratory Analyses
$$$ = ???
![Page 17: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/17.jpg)
Hanna Wallach :: UMass Amherst :: 18
Bayesian Latent Variable Models
● Modeling challenges:
– Aggregating and representing large data sets
– Handling data from sources with disparate emphases
– Efficiently reasoning under uncertain information
● Bayesian latent (i.e., hidden) variable models:
– Appropriate for prediction, explanation, and exploration
– Interpretable structure, not “black-box” models
– Powerful, flexible, widely applicable...
![Page 18: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/18.jpg)
Hanna Wallach :: UMass Amherst :: 19
This Talk
![Page 19: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/19.jpg)
Hanna Wallach :: UMass Amherst :: 20
This Talk
![Page 20: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/20.jpg)
Hanna Wallach :: UMass Amherst :: 21
Communication Networks
![Page 21: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/21.jpg)
Hanna Wallach :: UMass Amherst :: 22
Communication Networks
![Page 22: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/22.jpg)
Hanna Wallach :: UMass Amherst :: 23
Communication Networks
![Page 23: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/23.jpg)
Hanna Wallach :: UMass Amherst :: 24
Observing Communication Networks
![Page 24: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/24.jpg)
Hanna Wallach :: UMass Amherst :: 25
Structure and Content
Subject: New Hanover County Public Safety Talk GroupsFrom: “Lee, Warren” <[email protected]>To: “Pope, Troy W.” <[email protected]>Cc: …
Troy,
I wanted to give you anupdate on our progress inmoving towards a fullydigital public safety radiosystem in New HanoverCounty...
![Page 25: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/25.jpg)
Hanna Wallach :: UMass Amherst :: 26
New Hanover County, NC
![Page 26: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/26.jpg)
Hanna Wallach :: UMass Amherst :: 27
NHC Email Network
![Page 27: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/27.jpg)
Hanna Wallach :: UMass Amherst :: 28
Levels of Granularity
= ???
![Page 28: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/28.jpg)
Hanna Wallach :: UMass Amherst :: 29
Levels of Granularity
= ???
![Page 29: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/29.jpg)
Hanna Wallach :: UMass Amherst :: 30
Levels of Granularity
= ??? = ??? +
![Page 30: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/30.jpg)
Hanna Wallach :: UMass Amherst :: 31
Principled Visualization
● Common workflow:
– Construct a statistical model of observed data
– Perform post-hoc visualization to draw conclusions about the model and its relationship to the data
● Problem: visualization algorithms can produce visual artifacts that may be misleading
● Solution: visualizations should be directly interpretable in terms of the model and its relationship to the data
![Page 31: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/31.jpg)
Hanna Wallach :: UMass Amherst :: 32
Exploring Structure and Content
● Facilitate exploratory analysis of topic-specific communication patterns by learning
– Topics of communication
– Topic-specific communication subnetworks
– Principled visualizations of topic-sepcific subnetwork
● Draw upon ideas from two well-known frameworks:
– Statistical topic modeling
– Latent space network modeling
![Page 32: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/32.jpg)
Hanna Wallach :: UMass Amherst :: 33
Topics and Words
gene ncbi computer patent
genome national modeling patenting
dna information data claims
genetic technology algorithm intellectual
genes database analyses property
sequence molecular method rights
human biology model ip
protein genbank information innovation
rna pubmed efficient claim
genomic references complexity claiming
... ... ... ...
pro
babili
ty
![Page 33: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/33.jpg)
Hanna Wallach :: UMass Amherst :: 34
Documents and Topics
![Page 34: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/34.jpg)
Hanna Wallach :: UMass Amherst :: 35
Latent Dirichlet Allocation
probability
...
??
?? ?
[Blei, Ng & Jordan, '03]
![Page 35: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/35.jpg)
Hanna Wallach :: UMass Amherst :: 36
Individuals and Latent Spaces
every individual is associated
with a position in latent space
![Page 36: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/36.jpg)
Hanna Wallach :: UMass Amherst :: 37
Latent Space Network Model
probability of communication
depends on distance in
latent space
[Hoff et al., '02]
![Page 37: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/37.jpg)
Hanna Wallach :: UMass Amherst :: 38
Topics and Spaces
gene ncbi computer patent
genome national modeling patenting
dna information data claims
genetic technology algorithm intellectual
... ... ... ...
9
456
12
3
710
8 2
109
8
43
65
7
1210
9
84
3
65
71
2
98 10
14
3
56
7
![Page 38: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/38.jpg)
Hanna Wallach :: UMass Amherst :: 39
A New Model...
● Model email content using LDA
● Model recipients using topic-specific latent spaces
● Generative process:
– Generate topics and topic-specific latent spaces
– Generate document-specific topic distributions
– Generate recipients using latent spaces
– Generate words using topics
[Krafft et al., '12]
![Page 39: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/39.jpg)
Hanna Wallach :: UMass Amherst :: 40
Graphical Model
topicinference
subnetworkinference
subnetworkvisualization
![Page 40: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/40.jpg)
Hanna Wallach :: UMass Amherst :: 41
Experimental Evaluation
● Quantitative model validation:
– Link prediction performance vs. baselines
– Posterior predictive checks
– Topic coherence vs. LDA
● Exploratory analysis:
– Modularity: disconnected components
– Assortativity: components of a single “type”
![Page 41: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/41.jpg)
Hanna Wallach :: UMass Amherst :: 42
Link Prediction
![Page 42: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/42.jpg)
Hanna Wallach :: UMass Amherst :: 43
Posterior Predictive Checks
degree geodesic distance
transitivity dyadic intensity
![Page 43: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/43.jpg)
Hanna Wallach :: UMass Amherst :: 44
Topic Coherence
[Mimno et al., '11]
![Page 44: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/44.jpg)
Hanna Wallach :: UMass Amherst :: 45
Organization Structure
![Page 45: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/45.jpg)
Hanna Wallach :: UMass Amherst :: 46
High Modularity, High Assortativity
Meeting Schedulingmeeting march board agenda week
![Page 46: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/46.jpg)
Hanna Wallach :: UMass Amherst :: 47
High Modularity, Low Assortativity
Public Signagechange signs sign process ordinance
![Page 47: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/47.jpg)
Hanna Wallach :: UMass Amherst :: 48
Low Modularity, Low Assortativity
Public Relationscity breakdown information give
![Page 48: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/48.jpg)
Hanna Wallach :: UMass Amherst :: 49
Low Modularity, High Assortativity
Broadcast Messagesfw fyi bulletin summary week
![Page 49: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/49.jpg)
Hanna Wallach :: UMass Amherst :: 50
Take Away Message
● Explanatory and exploratory analyses matter
● Communication networks are important:
– Critical to all kinds of collaborative problem solving
– … but can be hard to directly observe
● Topic-partitioned multinetwork embedding:
– Good model of structure and content
– Emphasizes principled visualization
![Page 50: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/50.jpg)
Hanna Wallach :: UMass Amherst :: 51
This Talk
![Page 51: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/51.jpg)
Hanna Wallach :: UMass Amherst :: 52
Transparency in the US
● 52.8 million pages reviewed for declassification
● 26.7 million pages declassified
● $11.36 billion spent on administration of the US government classification system
[ISOO, 2011]
![Page 52: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/52.jpg)
Hanna Wallach :: UMass Amherst :: 53
● Date issued
● Date declassified
● Document type
● Source institution
● Classification level
● Document text
Declassified Documents
[Gale, 2012]
![Page 53: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/53.jpg)
Hanna Wallach :: UMass Amherst :: 54
Inferred Topics
% o
f avera
ge d
ocu
men
t
church
catholic
pope
vatican
religious
bishop
cardinal
archbishop
priests
paul
...
![Page 54: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/54.jpg)
Hanna Wallach :: UMass Amherst :: 55
draft
service
manpower
volunteer
selective
age
calls
volunteers
deferments
pay
...
Inferred Topics
% o
f avera
ge d
ocu
men
t
![Page 55: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/55.jpg)
Hanna Wallach :: UMass Amherst :: 56
atomic
weapon
bomb
bombs
weapon
energy
thermonuclear
development
hydrogen
stockpile
...
Inferred Topics
% o
f avera
ge d
ocu
men
t
![Page 56: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/56.jpg)
Hanna Wallach :: UMass Amherst :: 57
Predictive Analyses
$$$ ???
![Page 57: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/57.jpg)
Hanna Wallach :: UMass Amherst :: 58
Classification Duration
declassification date4/6/93
creation date2/21/67 26 years
![Page 58: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/58.jpg)
Hanna Wallach :: UMass Amherst :: 59
Survival Analysis
● Statistical methods for modeling durations:
– Biology/medicine: organism death
– Engineering: component failure
– Social sciences: event durations (e.g., recidivism)
● Goal: model effect on survival time of covariates, e.g.,
– Vaccine treatments
– Temperature differences
– Job placement or education programs
![Page 59: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/59.jpg)
Hanna Wallach :: UMass Amherst :: 60
Duration and Content
14 years 57 years
![Page 60: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/60.jpg)
Hanna Wallach :: UMass Amherst :: 61
● Topics provide information about classification durations
● Goal: incorporate durations into the probabilistic model
● Infer latent topics using both textual and temporal information
Modeling Text and Duration
![Page 61: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/61.jpg)
Hanna Wallach :: UMass Amherst :: 62
To Conclude...
![Page 62: Machine Learning for Complex Social Processeswallach/talks/2013-07-03_MSR_NE.pdf · Hanna Wallach :: UMass Amherst :: 18 Bayesian Latent Variable Models Modeling challenges: – Aggregating](https://reader033.fdocuments.in/reader033/viewer/2022042321/5f0bbb147e708231d431efe5/html5/thumbnails/62.jpg)
Thanks!
Acknowledgements: P. Krafft, J. Moore, B. Desmarais, R. Shorey
[email protected] http://www.cs.umass.edu/~wallach/