Analysis of Dynamics of

naivenorthAI and Robotics

Nov 8, 2013 (3 years and 7 months ago)

53 views

Analysis of Dynamics of
Research Communities

Gerd Stumme, Claudia Müller,
Vlado Batagelj, Bettina Hoser,
Steffen Staab

Application Questions



Create a new community or join an existing one?

(recommendations for community membership)



Discovery of convergence of research topics (M&A)



What are the main trends in SNA, ....? (relevance of
topics by relevant people)



Interdependencies/Co
-
evolution of “groups”

Generation of social and homophily networks

things

Homophily networks

Social networks

Groups

explicit / implicit

Groups

1,2,3,… mode networks

=?

actors

Data Flow

tags, words,
keywords,
events,
conferences

Similarity based on tags,

content, co
-
citation, co
-
visit

Communication networks

(email, co
-
author,…)

SNA community, SemWeb community

explicit (mailing list…) / implicit

Groups

Folksonomy

=?

Researchers,
institutions

Data Clustering Result

analyse or
recommend

Graph

Clustering

Result

Data Sets on Research Communities


Web of Science



Citeseer



ACM
-
Portal



DBLP



Bibsonomy


Research Questions



predictions from one network



Co
-
Citation network generates co
-
author network?



What citation structure generates co
-
author networks?

(assumption: bi
-
directional citation leads to co
-
authorship)



Does co
-
attendance of a Dagstuhl seminar generate co
-
author
-
ship?
(control group of declined invitations)

Research Questions



predictions from mixed networks



Co
-
author plus co
-
citation (or other homophily)
generates co
-
authorship?



Co
-
author plus co
-
citation recommends papers
(or authors)


Measures

things

Homophily networks

Social networks

Groups

(explicit / implicit)

groups

1,2,3,… mode networks

=?

actors

Jaccard
-
measure,…

Comparing
clusterings

Cosine,…

Graph
-
based
clustering

Data clustering,…

Time Modeling


Windowing (time slices)


(non
-
)overlapping windows?


Length of windows?


Different time
-
scales of different networks?


Decay of relationships?



Prediction:


Markovian vs non
-
Markovian (with memory)

Evolution of Groups

T
1

T
2

T
3

Evaluation

„Difficult tasks“:


Deviation from null hypothesis


Recommender tasks:


Predict the year 2007 from the time before

Session 2

Sociology vs Computer Science


Understand the
society under
consideration


Predict useful values
for your system

Evolution of Groups

T
1

T
2

T
3

Social
network A

Homophily
network B

Compare and
predict

Evolution of Groups

T
1

T
2

T
3

Social
network A

Homophily
network B

Compare &
predict