Abstract
Automated analysis of news reports is a significant empowering technology for predictive models of political instability. To date, the standard approach to this analytic task has been embodied in systems such as KEDS/TABARI [1], which use manually-generated rules and shallow parsing techniques to identify events and their participants in text. In this chapter we explore an alternative to event extraction based on BBN SERIFTM, and BBN OnTopicTM, two state-of-the-art statistical natural language processing engines. We empirically compare this new approach to existing event extraction techniques on five dimensions: (1) Accuracy: when an event is reported by the system, how often is it correct? (2) Coverage: how many events are correctly reported by the system? (3) Filtering of historical events: how well are historical events (e.g. 9/11) correctly filtered out of the current event data stream? (4) Topic-based event filtering: how well do systems filter out red herrings based on document topic, such as sports documents mentioning “clashes” between two countries on the playing field? (5) Domain shift: how well do event extraction models perform on data originating from diverse sources? In all dimensions we show significant improvement to the state-of-the-art by applying statistical natural language processing techniques. It is our hope that these results will lead to greater acceptance of automated coding by creators and consumers of social science models that depend on event data and provide a new way to improve the accuracy of those predictive models.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
- 2.
The evaluation corpus included approximately 250,000 documents. Documents judged to be a near duplicate via BBN’s semantic de-duplication filter were removed before evaluation.
- 3.
Specifically, the percentages of triples seen by two annotators were 100% for Violence, 100% for Provide Aid, and 69% for Disapprove.
- 4.
King and Lowe report a suite of numbers for accuracy; this number assumes a constant weighting across event categories. In addition, King and Lowe report an 85% accuracy number for a very different metric: the probability of a correct event or non-event judgment on a given sentence. We did not compute this number. Given the high percentage of non-event sentences in our data, it would be meaninglessly high—the trivial baseline, where a system never returns an event, would achieve 96% accuracy on our data set. In contrast, the King and Lowe test set is specifically constructed to contain mostly sentences that have a valid event of some kind (and their raw data pool is also more event-heavy than ours), so that number has a very different meaning in their context.
References
Schrodt P (2001) Automated coding of international event data using sparse parsing techniques. Paper presented at the International Studies Association, Chicago
O’Brien S (2010) Crisis early warning and decision support: contemporary approaches and thoughts on future research. Int Stud Rev 12:87–104. doi: 10.1111/j.1468-2486.2009.00914.x
Ramshaw L, Boschee E, Freedman M, MacBride J, Weischedel R, Zamanian A (2011) SERIF language processing—effective trainable language understanding. Handbook of natural language processing and machine translation: DARPA Global Autonomous Language Exploitation. Springer, New York
Gerner D, Schrodt P, Yilmaz O, Abu-Jabr R (2002) Conflict and Mediation Event Observations (CAMEO): a new event data framework for the analysis of foreign policy interactions. Paper presented at the International Studies Association, New Orleans, and American Political Science Association, Boston
Schrodt P, Yilmaz O, Gerner D, Hermreck D (2008) The CAMEO (Conflict and Mediation Event Observations) actor coding framework. Paper presented at the International Studies Association, San Francisco
Olive J, Christianson C, McCary J (2011) Handbook of natural language processing and machine translation: DARPA Global Autonomous Language Exploitation. Springer, New York
Maybury M (2004) New directions in question answering. AAAI Press/The MIT Press, Menlo Park
King G, Lowe W (2003) An automated information extraction tool for international conflict data with performance as good as human coders: a rare events evaluation design. Int Organ 57:617–642
Schwartz R, Imai T, Kubala F, Nguyen L, and Makhoul J (1997) A maximum likelihood model for topic classification of broadcast news. Proceedings of Eurospeech, Greece
Prasad R, Natarajan P, Subramanian K, Saleem S, Schwartz R (2007) Finding structure in noisy text: topic classification and unsupervised clustering. Paper presented at IJCAI-2007 Workshop on Analytics for Noisy Unstructured Text Data, Hyderabad, India
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2013 Springer Science+Business Media New York
About this chapter
Cite this chapter
Boschee, E., Natarajan, P., Weischedel, R. (2013). Automatic Extraction of Events from Open Source Text for Predictive Forecasting. In: Subrahmanian, V. (eds) Handbook of Computational Approaches to Counterterrorism. Springer, New York, NY. https://doi.org/10.1007/978-1-4614-5311-6_3
Download citation
DOI: https://doi.org/10.1007/978-1-4614-5311-6_3
Published:
Publisher Name: Springer, New York, NY
Print ISBN: 978-1-4614-5310-9
Online ISBN: 978-1-4614-5311-6
eBook Packages: Computer ScienceComputer Science (R0)