Link prediction in relational data

DMCA

Link prediction in relational data (2003)

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Ben Taskar , Ming-fai Wong , Pieter Abbeel , Daphne Koller

Venue:	in Neural Information Processing Systems
Citations:	156 - 1 self

BibTeX

@INPROCEEDINGS{Taskar03linkprediction,
    author = {Ben Taskar and Ming-fai Wong and Pieter Abbeel and Daphne Koller},
    title = {Link prediction in relational data},
    booktitle = {in Neural Information Processing Systems},
    year = {2003}
}

OpenURL

Abstract

Many real-world domains are relational in nature, consisting of a set of objects related to each other in complex ways. This paper focuses on predicting the existence and the type of links between entities in such domains. We apply the relational Markov network framework of Taskar et al. to define a joint probabilistic model over the entire link graph — entity attributes and links. The application of the RMN algorithm to this task requires the definition of probabilistic patterns over subgraph structures. We apply this method to two new relational datasets, one involving university webpages, and the other a social network. We show that the collective classification approach of RMNs, and the introduction of subgraph patterns over link labels, provide significant improvements in accuracy over flat classification, which attempts to predict each link in isolation. 1

Keyphrases

link prediction relational data flat classification link label probabilistic pattern subgraph pattern new relational datasets entire link graph entity attribute complex way significant improvement social network collective classification approach relational markov network framework rmn algorithm many real-world domain joint probabilistic model university webpage subgraph structure

DMCA