Title
A Semantic Feature For Verbal Predicate And Semantic Role Labeling Using Svms
Abstract
This paper shows that semantic role labeling is a consequence of accurate verbal predicate labeling. In doing so, the paper presents a novel type of semantic feature for verbal predicate labeling using a new corpus. The corpus contains verbal predicates, serving as verb senses, that have semantic roles associated with each argument. Although much work has been done using feature vectors with machine learning algorithms for various types of semantic classification tasks, past work has primarily shown effective use of syntactic or lexical information. Our new type of semantic feature, ontological regions, proves highly effective when used in addition to or in place of syntactic and lexical features for support vector classification, increasing accuracy of verbal predicate labeling from 65.4% to 78.8%. Copyright © 2008, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.
Publication Date
11-17-2008
Publication Title
Proceedings of the 21th International Florida Artificial Intelligence Research Society Conference, FLAIRS-21
Number of Pages
213-218
Document Type
Article; Proceedings Paper
Personal Identifier
scopus
Copyright Status
Unknown
Socpus ID
55849095813 (Scopus)
Source API URL
https://api.elsevier.com/content/abstract/scopus_id/55849095813
STARS Citation
Schwartz, Hansen A.; Gomez, Fernando; and Millward, Christopher, "A Semantic Feature For Verbal Predicate And Semantic Role Labeling Using Svms" (2008). Scopus Export 2000s. 9715.
https://stars.library.ucf.edu/scopus2000/9715