Social tagging in the life sciences: characterizing a new metadata resource for bioinformatics

Area of Science:

• Life Sciences
• Bibliometrics
• Information Science

Background:

• Academic social tagging systems (e.g., Connotea, CiteULike) generate metadata repositories from user-generated tags.
• These repositories are a potential resource for life sciences information management and application development.
• The study characterizes metadata from two prominent social tagging systems.

Purpose of the Study:

• To assess the metadata quality and characteristics of social tagging systems.
• To determine the potential utility of socially constructed metadata for life sciences.
• To suggest improvements for future social tagging system design.

Main Methods:

• Evaluated metadata from CiteULike and Connotea for citations in PubMed.
• Assessed metrics including document coverage, tag density, inter-annotator agreement, and agreement with Medical Subject Headings (MeSH).

Main Results:

• CiteULike and Connotea exhibited similar metadata characteristics.
• Social tagging systems had lower document coverage and tag density compared to PubMed.
• Inter-annotator agreement and agreement with MeSH indexing were low, but could be improved with voting mechanisms.

Conclusions:

• Novel uses of social tagging metadata may leverage user-tag-document relationships.
• Traditional indexing applications require increased tagging and user participation for substantial benefit.
• Future systems should focus on attracting users and designing interfaces that promote useful tagging behavior.