Previous work in social network analysis (SNA)
has modeled the existence of links from one entity
to another, but not the language content or
topics on those links. We present the Author-
Recipient-Topic (ART) model for social network
analysis, which learns topic distributions based on
the direction-sensitive messages sent between entities.
The model builds on Latent Dirichlet Allocation
(LDA) and the Author-Topic (AT) model,
adding the key attribute that distribution over topics
is conditioned distinctly on both the sender and
recipient—steering the discovery of topics according
to the relationships between people. We give
results on both the Enron email corpus and a researcher’s
email archive, providing evidence not
only that clearly relevant topics are discovered, but
that the ART model better predicts people’s roles.
%0 Conference Paper
%1 citeulike:344452
%A Mccallum, A.
%A Corrada-Emmanuel, A.
%A Wang, X.
%B ijcai.org
%D 2005
%K lda network role social topic
%T Topic and role discovery in social networks
%U http://www.cs.umass.edu/~mccallum/papers/art-ijcai05.pdf
%X Previous work in social network analysis (SNA)
has modeled the existence of links from one entity
to another, but not the language content or
topics on those links. We present the Author-
Recipient-Topic (ART) model for social network
analysis, which learns topic distributions based on
the direction-sensitive messages sent between entities.
The model builds on Latent Dirichlet Allocation
(LDA) and the Author-Topic (AT) model,
adding the key attribute that distribution over topics
is conditioned distinctly on both the sender and
recipient—steering the discovery of topics according
to the relationships between people. We give
results on both the Enron email corpus and a researcher’s
email archive, providing evidence not
only that clearly relevant topics are discovered, but
that the ART model better predicts people’s roles.
@inproceedings{citeulike:344452,
abstract = {Previous work in social network analysis (SNA)
has modeled the existence of links from one entity
to another, but not the language content or
topics on those links. We present the Author-
Recipient-Topic (ART) model for social network
analysis, which learns topic distributions based on
the direction-sensitive messages sent between entities.
The model builds on Latent Dirichlet Allocation
(LDA) and the Author-Topic (AT) model,
adding the key attribute that distribution over topics
is conditioned distinctly on both the sender and
recipient—steering the discovery of topics according
to the relationships between people. We give
results on both the Enron email corpus and a researcher’s
email archive, providing evidence not
only that clearly relevant topics are discovered, but
that the ART model better predicts people’s roles.},
added-at = {2007-08-15T12:03:25.000+0200},
author = {Mccallum, A. and Corrada-Emmanuel, A. and Wang, X.},
biburl = {https://www.bibsonomy.org/bibtex/24a5138f8d572d2f89e2b94ec60986278/wnpxrz},
booktitle = {ijcai.org},
citeulike-article-id = {344452},
comment = {sent by Thomas Hofmann
---
Domain: email (enron + private)
Task: discovering topics, clustering to find social roles, summarizing,
similar roles (automated coreference system, expert-finding, message recommendation)
Method: Author-Recipient-Topic (ART) resp. Role-Author-Recipient-Topic (RART). Based on Bayesian Networks. Predecessors: Latent Dirichlet Allocation (Blei, Ng, Jordan, 2003), Author-Model (McCallum 99), Author-Topic-Model (Rosen-Zvi, Griffith, Steyvers, Smyth, 04)
Motto: "distinguish author and recipient of a message"},
interhash = {2f4fff68939c9125ce6f357ab7092223},
intrahash = {4a5138f8d572d2f89e2b94ec60986278},
keywords = {lda network role social topic},
priority = {0},
timestamp = {2007-08-15T12:03:25.000+0200},
title = {Topic and role discovery in social networks},
url = {http://www.cs.umass.edu/~mccallum/papers/art-ijcai05.pdf},
year = 2005
}