Leveraging Relational Autocorrelation
with Latent Group Models
J. Neville, and D. Jensen. Proceedings of the 4th Multi-Relational Data Mining Workshop, 11th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, (2005)
Abstract
The presence of autocorrelation provides strong motivation for using
relational techniques for learning and inference. Autocorrelation
is a statistical dependency between the values of the same variable
on related entities and is a nearly ubiquitous characteristic of
relational data sets. Recent research has explored the use of collective
inference techniques to exploit this phenomenon. These techniques
achieve significant performance gains by modeling observed
correlations among class labels of related instances, but the models
fail to capture a frequent cause of autocorrelation—the presence
of underlying groups that influence the attributes on a set of entities.
We propose a latent group model (LGM) for relational data,
which discovers and exploits the hidden structures responsible for
the observed autocorrelation among class labels. Modeling the latent
group structure improves model performance, increases inference
efficiency, and enhances our understanding of the datasets.
We evaluate performance on three relational classification tasks and
show that LGM outperforms models that ignore latent group structure,
particularly when there is little information with which to seed
inference.
%0 Conference Paper
%1 citeulike:394156
%A Neville, Jennifer
%A Jensen, David
%B Proceedings of the 4th Multi-Relational Data Mining Workshop, 11th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining
%D 2005
%K topicinference groupmodel socialnets
%T Leveraging Relational Autocorrelation
with Latent Group Models
%U http://kdl.cs.umass.edu/papers/neville-jensen-mrdm2005.pdf
%X The presence of autocorrelation provides strong motivation for using
relational techniques for learning and inference. Autocorrelation
is a statistical dependency between the values of the same variable
on related entities and is a nearly ubiquitous characteristic of
relational data sets. Recent research has explored the use of collective
inference techniques to exploit this phenomenon. These techniques
achieve significant performance gains by modeling observed
correlations among class labels of related instances, but the models
fail to capture a frequent cause of autocorrelation—the presence
of underlying groups that influence the attributes on a set of entities.
We propose a latent group model (LGM) for relational data,
which discovers and exploits the hidden structures responsible for
the observed autocorrelation among class labels. Modeling the latent
group structure improves model performance, increases inference
efficiency, and enhances our understanding of the datasets.
We evaluate performance on three relational classification tasks and
show that LGM outperforms models that ignore latent group structure,
particularly when there is little information with which to seed
inference.
@inproceedings{citeulike:394156,
abstract = {The presence of autocorrelation provides strong motivation for using
relational techniques for learning and inference. Autocorrelation
is a statistical dependency between the values of the same variable
on related entities and is a nearly ubiquitous characteristic of
relational data sets. Recent research has explored the use of collective
inference techniques to exploit this phenomenon. These techniques
achieve significant performance gains by modeling observed
correlations among class labels of related instances, but the models
fail to capture a frequent cause of autocorrelation—the presence
of underlying groups that influence the attributes on a set of entities.
We propose a latent group model (LGM) for relational data,
which discovers and exploits the hidden structures responsible for
the observed autocorrelation among class labels. Modeling the latent
group structure improves model performance, increases inference
efficiency, and enhances our understanding of the datasets.
We evaluate performance on three relational classification tasks and
show that LGM outperforms models that ignore latent group structure,
particularly when there is little information with which to seed
inference.},
added-at = {2006-06-16T10:34:37.000+0200},
author = {Neville, Jennifer and Jensen, David},
biburl = {https://www.bibsonomy.org/bibtex/24d6907df6fbfdd6effeab2a6e7a60cb8/ldietz},
booktitle = {Proceedings of the 4th Multi-Relational Data Mining Workshop, 11th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining},
citeulike-article-id = {394156},
comment = {Similar to http://www.citeulike.org/user/ldietz/article/397080},
interhash = {f8dfa62109d4db16599174a372fbb207},
intrahash = {4d6907df6fbfdd6effeab2a6e7a60cb8},
keywords = {topicinference groupmodel socialnets},
priority = {2},
timestamp = {2006-06-16T10:34:37.000+0200},
title = {Leveraging Relational Autocorrelation
with Latent Group Models},
url = {http://kdl.cs.umass.edu/papers/neville-jensen-mrdm2005.pdf},
year = 2005
}