Engineering Village
Search History  -  Selected Records  -  My Profile  -  My Alerts End Session
Tag Search Easy Search Quick Search Expert Search Thesaurus Ask an Expert Help
Abstract  -  Detailed                
Record 1 from Compendex for: ((Applying Machine Learning to Chinese Entity Detection and Tracking) WN All fields), 1899-2008
Check record to add to Selected Records
Accession number: 075110984085
Title: Applying machine learning to chinese entity detection and tracking
Authors: Qian, Donglei; Li, Wenjie; Yuan, Chunfa; Lu, Qin; Wu, Mingli
Author affiliation: Department of Computing, Hong Kong Polytechnic University, Hong Kong, Hong Kong
Serial title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Abbreviated serial title: Lect. Notes Comput. Sci.
Volume: v 4394 LNCS
Monograph title: Computational Linguistics and Intelligent Text Processing - 8th International Conference, CICLing 2007, Proceedings
Publication year: 2007
Pages: p 154-165
Language: English
ISSN: 0302-9743
Document type: Conference article (CA)
Conference name: 8th Annual Conference on Intelligent Text Processing and Computational Linguistics, CICLing 2007
Conference date: Feb 18-24 2007
Conference location: Mexico City, Mexico
Conference code: 70754
Publisher: Springer Verlag, Heidelberg, D-69121, Germany
Abstract: This paper presents a Chinese entity detection and tracking system that takes advantages of character-based models and machine learning approaches. An entity here is defined as a link of all its mentions in text together with the associated attributes. Entity mentions of different types normally exhibit quite different linguistic patterns. Six separate Conditional Random Fields (CRF) models that incorporate character N-gram and word knowledge features are built to detect the extent and the head of three types of mentions, namely named, nominal and pronominal mentions. For each type of mentions, attributes are identified by Support Vector Machine (SVM) classifiers which take mention heads and their context as classification features. Mentions can then be merged into a unified entity representation by examining their attributes and connections in a rule-based coreference resolution process. The system is evaluated on ACE 2005 corpus and achieves competitive results. © Springer-Verlag Berlin Heidelberg 2007.
Number of references: 18
Ei main heading: Support vector machines
Ei controlled terms: Classification (of information)  -  Feature extraction  -  Knowledge acquisition  -  Target tracking
Uncontrolled terms: Conditional Random Fields (CRF)  -  Knowledge features  -  Chinese entity detection
Ei classification codes: 716.1 Information Theory and Signal Processing  -  716.2 Radar Systems and Equipment  -  723 Computer Software, Data Handling and Applications  -  723.4 Artificial Intelligence  -  723.5 Computer Applications  -  903.1 Information Sources and Analysis
Treatment: Theoretical (THR)
Database: Compendex
Compilation and indexing terms, © 2008 Elsevier Inc.
 Separate each tag with a comma: nanotech, carbon nanotubes, Environmental 101

  Full-text and Local Holdings Links

About Ei  -  About Engineering Village  -  Feedback  -  Privacy Policy  -  Terms and Conditions
© 2008 Elsevier Inc. All rights reserved.