Automatically identifying tag types

authored by: Kerstin Bischoff, Claudiu S. Firan, Cristina Kadar, Wolfgang Nejdl, Raluca Paiu
Abstract: Web 2.0 applications such as delicious, flickr or lastfm have recently become extremely popular and as a result, a large amount of semantically rich metadata produced by users becomes available and exploitable. Tag information can be used for many purposes (e.g. user profiling, recommendations, clustering etc), though the benefit of tags for search is by far the most discussed usage. Tag types differ largely across systems and previous studies showed that, while some tag type categories might be useful for some particular users when searching, they may not bring any benefit to others. The present paper proposes an approach which utilizes rule-based as well as model-based methods, in order to automatically identify exactly these different types of tags. We compare the automatic tag classification produced by our algorithms against a ground truth data set, consisting of manual tag type assignments produced by human raters. Experimental results show that our methods can identify tag types with high accuracy, thus enabling further improvement of systems making use of social tags.
Organisation(s): L3S Research Centre
Type: Conference contribution
Pages: 31-42
No. of pages: 12
Publication date: 2009
Publication status: Published
Peer reviewed: Yes
ASJC Scopus subject areas: Theoretical Computer Science, General Computer Science
Electronic version(s): https://doi.org/10.1007/978-3-642-03348-3_7 (Access: Closed)

BibTeX

@inproceedings{8802978fbd1c4b33b94ad3c2e01256dc,
title = "Automatically identifying tag types",
abstract = "Web 2.0 applications such as delicious, flickr or lastfm have recently become extremely popular and as a result, a large amount of semantically rich metadata produced by users becomes available and exploitable. Tag information can be used for many purposes (e.g. user profiling, recommendations, clustering etc), though the benefit of tags for search is by far the most discussed usage. Tag types differ largely across systems and previous studies showed that, while some tag type categories might be useful for some particular users when searching, they may not bring any benefit to others. The present paper proposes an approach which utilizes rule-based as well as model-based methods, in order to automatically identify exactly these different types of tags. We compare the automatic tag classification produced by our algorithms against a ground truth data set, consisting of manual tag type assignments produced by human raters. Experimental results show that our methods can identify tag types with high accuracy, thus enabling further improvement of systems making use of social tags.",
keywords = "Classification, Collaborative tagging, Social media, Tag types",
author = "Kerstin Bischoff and Firan, {Claudiu S.} and Cristina Kadar and Wolfgang Nejdl and Raluca Paiu",
year = "2009",
doi = "10.1007/978-3-642-03348-3_7",
language = "English",
isbn = "978-3-642-03347-6",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "31--42",
booktitle = "Advanced Data Mining and Applications",
note = "5th International Conference on Advanced Data Mining and Applications, ADMA 2009 ; Conference date: 17-08-2009 Through 19-08-2009",
}