Topic Detection and Tracking: Event Clustering as a Basis for First Story Detection

Papka, Ron; Allan, James

doi:10.1007/0-306-47019-5_4

Cited by 16 publications

(21 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For instance, Acts of Congress during the Civil War could be compared to military events in the War. An event gazetteer could be the basis of for browsing newspapers stories (e.g., [3,7]). It could also manage events presented as part of scientific explanations.…”

Section: The Visionmentioning

confidence: 99%

A query interface for an event gazetteer

Allen

2004

Proceedings of the 4th ACM/IEEE-CS Joint Conference on Digital Libraries

View full text Add to dashboard Cite

We introduce an event gazetteer which stores and presents "locations in time". Each event is coded with attributes of event type, location, actor, and beginning and ending times. Events can also contain sets of other events. This paper reports the development of an interface for generating searches to these "part-of" relationships. For instance, we can search for all named Battles in the event database which occurred during the Civil War. Ultimately, we envision a flexible, broad-based service that is a resource for users ranging from students to genealogists and researchers interested in historical events.

show abstract

Section: The Visionmentioning

confidence: 99%

A query interface for an event gazetteer

Allen

2004

Proceedings of the 4th ACM/IEEE-CS Joint Conference on Digital Libraries

View full text Add to dashboard Cite

show abstract

“…Basically, a sentence is compared to all previous sentences, and if the similarities to all of the previous sentences are below a threshold, the sentence is labeled as novel. This idea is adapted from FSD in TDT (Papka, 1999). M.‐F.…”

Section: Related Workmentioning

confidence: 99%

Novelty detection for topic tracking

Aksoy

Can

Kocberber

2011

Acta Anaesthesiol Scand

View full text Add to dashboard Cite

Multisource web news portals provide various advantages such as richness in news content and an opportunity to follow developments from different perspectives. However, in such environments, news variety and quantity can have an overwhelming effect. New-event detection and topic-tracking studies address this problem. They examine news streams and organize stories according to their events; however, several tracking stories of an event/topic may contain no new information (i.e., no novelty). We study the novelty detection (ND) problem on the tracking news of a particular topic. For this purpose, we build a Turkish ND test collection called BilNov-2005 and propose the usage of three ND methods: a cosine-similarity (CS)-based method, a language-model (LM)-based method, and a cover-coefficient (CC)-based method. For the LM-based ND method, we show that a simpler smoothing approach, Dirichlet smoothing, can have similar performance to a more complex smoothing approach, Shrinkage smoothing. We introduce a baseline that shows the performance of a system with random novelty decisions. In addition, a category-based threshold learning method is used for the first time in ND literature. The experimental results show that the LM-based ND method significantly outperforms the CS-and CC-based methods, and categorybased threshold learning achieves promising results when compared to general threshold learning. IntroductionThe Internet has changed the news industry (The Economist, 2011). Most newspapers and news agencies provide news on their web pages. News portals work as a news aggregator and gather, merge, and organize news articles obtained from various sources. Multisource news portals provide various advantages such as richness in news content and an opportunity to follow event developments from different perspectives. In addition, it is practical to Received May 20, 2011; revised September 12, 2011; accepted September 28, 2011 1 Present address: Computer Science Department, New Jersey Institute of Technology, University Heights Newark, NJ 07102.© 2011 ASIS&T • Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/asi.21697 follow different news sources from a single web page. Google News (http://news.google.com) is a well-known commercial news portal example. It offers many services such as information retrieval, personalized information filtering, and news clustering. Research-oriented examples include NewsBlaster (McKeown et al., 2002) and NewsInEssence (Radev, Otterbacher, Winkel, & Blair-Goldensohn, 2005), each of which provides clustering and summarization services over the news.As the number of sources and events increase, news readers may be overloaded with information and thus may face difficulty in finding news related to their interests. Different organizational techniques have been employed for more effective, efficient, and enjoyable browsing. Studies on new-event detection and topic tracking aim to organize news with respect to events or topics. In topic detection and trackin...

show abstract

“…If, on the other hand, the story is different enough from everything already seen, it is declared "new" and starts a new event in the history. Researchers have developed a range of techniques to address this problem, with the most typical being a form of vector-based document clusterings [28,30,31,32].…”

Section: New Event Detectionmentioning

confidence: 99%

Robust Techniques for Organizing and Retrieving Spoken Documents

Allan

2003

EURASIP J. Adv. Signal Process.

Self Cite

View full text Add to dashboard Cite

Information retrieval tasks such as document retrieval and topic detection and tracking (TDT) show little degradation when applied to speech recognizer output. We claim that the robustness of the process is because of inherent redundancy in the problem: not only are words repeated, but semantically related words also provide support. We show how document and query expansion can enhance that redundancy and make document retrieval robust to speech recognition errors. We show that the same effect is true for TDT′s tracking task, but that recognizer errors are more of an issue for new event and story link detection.

show abstract

Topic Detection and Tracking: Event Clustering as a Basis for First Story Detection

Cited by 16 publications

References 15 publications

A query interface for an event gazetteer

A query interface for an event gazetteer

Novelty detection for topic tracking

Robust Techniques for Organizing and Retrieving Spoken Documents

Contact Info

Product

Resources

About