J. Piskorski, H. Tanev, M. Atkinson, and E. Goot. Proceeding of the 2008 Conference on New Trends in Multimedia and Network Information Systems, page 276--290. IOS Press, (2008)
Please log in to take part in the discussion (add own reviews or comments).
Cite this publication
More citation styles
- please select -
%0 Conference Paper
%1 piskorski2008
%A Piskorski, Jakub
%A Tanev, Hristo
%A Atkinson, Martin
%A Goot, Erik Van Der
%B Proceeding of the 2008 Conference on New Trends in Multimedia and Network Information Systems
%D 2008
%I IOS Press
%K bootstrap cross-document cross-lingual dataset_newsfeed dataset_own event_extraction event_unification onlinedemo
%P 276--290
%T Cluster-Centric Approach to News Event Extraction
%@ 978-1-58603-904-2
@inproceedings{piskorski2008,
added-at = {2009-10-22T15:19:15.000+0200},
author = {Piskorski, Jakub and Tanev, Hristo and Atkinson, Martin and Goot, Erik Van Der},
biburl = {https://www.bibsonomy.org/bibtex/2753a4f254965c095e14695c05d231936/jnothman},
booktitle = {Proceeding of the 2008 Conference on New Trends in Multimedia and Network Information Systems},
description = {Multilingual event extraction system focussing on clustered first sentences of news articles.
Focus on violent and natural disaster events: date, location, number killed and injured, actors, kidnapped people, type of event.
Realtime hierarchical clustering of news (bag of words vector cosine, with some entropy-based stopwording, from the first 200 worsd). Clusters and individual articles are geolocated.
Surface extraction and role-labelling grammars (manually built?) with language and domain-specific lexicons.
Bootstrapped pattern acquisition with manual validation for each iteration.
Event unification:
- semantic roles merged according to reliability=type of extraction rules;
- victim counting: votes with numerical similarity estimation;
- event type classification: event types ranked according to occurrence of type-specific keywords in articles.
Evaluation not sufficiently detailed.},
interhash = {ec27bfe389f3cc182016c54f902daef9},
intrahash = {753a4f254965c095e14695c05d231936},
isbn = {978-1-58603-904-2},
keywords = {bootstrap cross-document cross-lingual dataset_newsfeed dataset_own event_extraction event_unification onlinedemo},
pages = {276--290},
publisher = {IOS Press},
timestamp = {2009-10-22T15:19:15.000+0200},
title = {Cluster-Centric Approach to News Event Extraction},
year = 2008
}