Exploring The Value Of Folksonomies For Creating Semantic Metadata
Exploring The Value Of Folksonomies For Creating Semantic Metadata
Finding good keywords to describe resources is an on-going problem: typically we select such words manually from a thesaurus of terms, or they are created using automatic keyword extraction techniques. Folksonomies are an increasingly well populated source of unstructured tags describing web resources. This paper explores the value of the folksonomy tags as potential source of keyword metadata by examining the relationship between folksonomies, community produced annotations, and keywords extracted by machines. The experiment has been carried-out in two ways: subjectively, by asking two human indexers to evaluate the quality of the generated keywords from both systems; and automatically, by measuring the percentage of overlap between the folksonomy set and machine generated keywords set. The results of this experiment show that the folksonomy tags agree more closely with the human generated keywords than those automatically generated. The results also showed that the trained indexers preferred the semantics of folksonomy tags compared to keywords extracted automatically. These results can be considered as evidence for the strong relationship of folksonomies to the human indexer’s mindset, demonstrating that folksonomies used in the del.icio.us bookmarking service are a potential source for generating semantic metadata to annotate web resources.
Metadata, Web Technologies, Folksonomy, Keyword Extraction, Tags, Social Bookmarking Services, Collaborative Tagging
13-39
Al-Khalifa, Hend S.
b464bced-9859-47ef-829a-4675dfdba01e
Davis, Hugh C.
1608a3c8-0920-4a0c-82b3-ee29a52e7c1b
March 2007
Al-Khalifa, Hend S.
b464bced-9859-47ef-829a-4675dfdba01e
Davis, Hugh C.
1608a3c8-0920-4a0c-82b3-ee29a52e7c1b
Al-Khalifa, Hend S. and Davis, Hugh C.
(2007)
Exploring The Value Of Folksonomies For Creating Semantic Metadata.
International Journal on Semantic Web and Information Systems (IJSWIS), 3 (1), .
Abstract
Finding good keywords to describe resources is an on-going problem: typically we select such words manually from a thesaurus of terms, or they are created using automatic keyword extraction techniques. Folksonomies are an increasingly well populated source of unstructured tags describing web resources. This paper explores the value of the folksonomy tags as potential source of keyword metadata by examining the relationship between folksonomies, community produced annotations, and keywords extracted by machines. The experiment has been carried-out in two ways: subjectively, by asking two human indexers to evaluate the quality of the generated keywords from both systems; and automatically, by measuring the percentage of overlap between the folksonomy set and machine generated keywords set. The results of this experiment show that the folksonomy tags agree more closely with the human generated keywords than those automatically generated. The results also showed that the trained indexers preferred the semantics of folksonomy tags compared to keywords extracted automatically. These results can be considered as evidence for the strong relationship of folksonomies to the human indexer’s mindset, demonstrating that folksonomies used in the del.icio.us bookmarking service are a potential source for generating semantic metadata to annotate web resources.
Text
Folksonomies.pdf
- Other
Text
IJSWIS_2007.pdf
- Other
More information
Published date: March 2007
Keywords:
Metadata, Web Technologies, Folksonomy, Keyword Extraction, Tags, Social Bookmarking Services, Collaborative Tagging
Organisations:
Web & Internet Science
Identifiers
Local EPrints ID: 263555
URI: http://eprints.soton.ac.uk/id/eprint/263555
PURE UUID: c179b966-0074-49be-8f72-e308577707b6
Catalogue record
Date deposited: 19 Feb 2007
Last modified: 15 Mar 2024 02:36
Export record
Contributors
Author:
Hend S. Al-Khalifa
Author:
Hugh C. Davis
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics