Tag! You’re it!

Article Review:

Fu, Y., Mao J., Su D., and Xu, Z. Towards the Semantic Web: Collaborative Tag Suggestions. Retrieved from: http://semanticmetadata.net/hosted/taggingws_www2006-files/13.pdf on February 2, 2012

Image retrieved from http://bit.ly/d2I6Dc

When it comes to organizing web objects, tagging has been a common method of classifying the object and identifying it for easy retrieval.  Tagging is a networked approach to organizing objects as opposed to conventional hierarchical structures.  Rather than a top-down cascading approach, tagging allows users to interact with objects as a network of data.

In browsing Google Scholar (which is my new favorite search engine for academic articles) I came across a very good article on using tagging as an effective way of organizing web objects.

The authors argue that tagging (as opposed to traditional methods of organizing data) allow users to move away from standard methods of fitting objects into  set-standard ‘molds’.  Tags used in combination with each other also allow for flexibility in searching.  The authors uses the example of Renewable Energy.  This tag allows users to search for items tagged under renewable, energy, and renewable energy.

Tagging allows for flexibility when searching for digital items.  Think about a physical photograph.  If the photograph is of a horse on the prairie,  which photo album would you put the photo in?  One of horses? One of the prairies? What about the time period?  What about a specific farm or location?  Digital tagging allows users to enter specific keywords that would draw a potential audience to that web object.  This is the case my team is experiencing with the Baker Slide collection.

When tagging an object, tags do not necessarily need to relate to each other.  The tags are meant to draw in the audience.  Think back to the horse on the prairie, ‘horse’ and ‘prairie’ are mutually exclusive to each other, however, they each draw an audience to the same photo.  The more keywords one searches, the more refined and specific the web object will be.

Xu, Fu, Mao and Su suggest that the number of tags be kept small, but fail to provide an ideal number.  In their research they used web objects that had between three and 6 tags.  They suggest using one tag from each of the categories below.  I will use the idea of the horse on the prairie as an example.

1) Content-based tags – These are tags that relate to what is contained in the photo.   They are specific terms.  These can be objects in the photo, or what the photo is about.   HORSE, WHEAT, etc.

2) Context-based tags – These tags are things which help identify the content.  They can be locations, places, and descriptors. PRAIRIE, ALBERTA

3) Attribute tags – These tags relate to the main idea, but may not be identified directly.  SMITH (for Joe Smith’s Horse), RAMSAY (as a descriptor of the land owner).

4) Subjective tags – These tags are how the creator feels about the images. PEACE, SERENE, COOL, etc.

5) Organizational tags – These tags help to organize, or define the organization to which the object belongs. Ex. BAKER SLIDES, etc.

The authors also set some criteria for good tags.  They say that tags should:

  • Include common words
  • High popularity
  • Identifying should be least-effort
  • Uniformity within a collections

In conclusion, Xu, Fu, Mao and Su believe that tags allow users to search for and reach web objects from multiple vectors (not a linear or uniform search). Tags allow for sharing and discovering web objects.  The authors also caution us to think and use tags appropriately and purposely to eliminate unsolicited searches and spam.  They also suggest that the same principles in using tags can be used in developing a ‘tag cloud’ – a grouping of tags to be used to tag a collection.