Future work on social tagging
The results from evaluations of social tags by experienced indexers in MELT highlighted a number of interesting issues that need further validation and investigation. Social tagging, as a feature in a conventional learning resource repository, is a very new phenomenon and it will take time before those interested in this approach have well developed evaluation methodologies and tools in this new context. Nevertheless, the MELT analysis shows that:
- Tags that expert indexers don't understand mostly constitute ‘noise’, but there are exceptions to this (see 2).
- Some tags travel across languages; i.e. people understand them even if they do not speak the language. These “travel well” tags can support retrieval in a multilingual context by facilitating the cross-border retrieval of resources.
- Some tags are understood only by a sub-group of users (e.g. “esl” = English as a Second Language) enhancing cross-border use and adding value for these sub-groups, but mostly they constitute ‘noise’ to others.
- Some tags correspond to descriptors in the LRE Thesaurus and can be used as indexing keywords for a resource, especially when the existing indexing is poor or the tag represents a narrower term.
- ”Thesaurus tags” can be used to determine the language equivalences between keywords, and affinities between tags and indexing keywords.
- Thesaurus terms could be used in order to determine affinities between tags, thus helping describe resources, as well as retrieval of resources in multiple languages.
- Tags can lead to interesting non-descriptors in the thesaurus and thus facilitate and enhance multilinguality.
- Tags can help enrich the thesaurus in terms of suggesting new descriptors based on how users have used tags to describe resources.
Lots of food for thought here that can be investigated further once a critical mass of user-generated tags has been accumulated as a result of many thousands of teachers using the public version of the LRE portal in 2009. Look out for a fast expanding LRE tag cloud!
The LRE tag cloud in May 2009