Library clips

sharing ideas thoughts and feedback

April 15, 2005

Social bookmarks so far…

Filed under: General, library, tags, folksonomy

The ultimate goal is for our social bookmark managers to be our personal search engines which can search full-text and also by a subject field (tag). As noted before there isn’t a tool yet that does both.

Added to this, is for the quality of the collective bookmarks to form an emerging vocabularly; a folksonomy that is scalable and usable(effective) - in essence a user defined subject index for the web.

…another issue gaining momentum is in capturing explicit metadata on user behaviour.

Folksonomy vocabulary

The authority on the subject term (tag) is the user, so for your personal collection the vocabularly is bound to make sense, but when the system combines everyone’s accounts and tags we have no single authority anymore, and the vocabularly no longer scales as well.

This is why in traditional systems, when the content grows, people decide it’s time to set rules, in essence, begin to control the vocabulary…these rules can be based on industry standards or home-made.

This is where the problem lies in folksonomies regarding social bookmark managers, there isn’t a controlled vocabularly (but that’s the whole idea) and they are trying to serve two things at once; the personal collection, and the collective collection…how do you have the best of both worlds!

Some obvious issues are synonym control, spelling mistakes, multiple terms for the same thing, same term for different things, etc…

…see these posts for more:

Folksonomies - Cooperative Classification and Communication Through Shared Metadata
A del.icio.us study - Bookmark, Classify and Share: A mini-ethnography of social practices in a distributed classification community.
Metadata for the masses
Folksonomy Notes: Considering the Downsides, Behavioral Trends, and Adaptation
Folksonomies: How we can improve the tags
Can social tagging overcome barriers to content classification?
Folksonomies? How about Metadata Ecologies?
folksonomies + controlled vocabularies
Controlled Vocabularies and Folksonomies: Why Change is Good.
Folksonomy Explanations
Folksonomy
Explaining and Showing Broad and Narrow Folksonomies
Taxonomies and Tags: From Trees to Piles of Leaves
A taxonomy of humour - what nurses can teach us about classification

The tags labels and the content within each tag may not be perfect or choice, or may contain errors (you may find bookmarks that don’t belong in a tag - but at the same time it’s in the eye of the indexer). The by-product of this is that it increases the chance of serendipity when compared to a controlled vocabularly (finding sites by accident is sometimes a gift from the gods).

The reason for this imperfection in indexing is that many people are doing the tagging (everyone has different ideas in judging the aboutness of something)…this problem exists even within a controlled vocabularly environment.
Also the tag index may lack conformity and organisation even within one account let alone collectively.

But even if there was a controlled vocabularly, how could it serve so many disciplines, this isn’t an easy task to map!…you’d need a meta-thesaurus…if the domain of the folksonomy was topic or discipline specific and serviced a particular group of people, eg. a company, then scaling the vocabulary would be not as a massive task.
An example of this on the open web is Connotea, a social bookmark manager (a free online reference management service for scientists) developed by Nature Publishing Group. Most of the members in this system are sharing similar interests, so the content in this domain is discipline specific (not strictly). In turn this makes it easier or more probable for a consistent vocabularly to emerge; a more workable folksonomy.

Another problem in building an emerging vocabularly of this sort is the dynamics of the content within peoples accounts.

If the content in my account is predominately about “Blogs”, I will not just tag items “Blogs”, I’ll get more specific and use tags like “Blogs_km” Blogs_edu” “Blogs_rsch”, etc…
If someone elses account has only a few bookmarks about “Blogs” they will more likely use the broad term “Blogs”.
Also if an account is topic specific, like the former, using tags requires unique names like “Blogs_km” which counters the emerging folksonomy. A way around this is using only one word for a tag and using the adding function to retrieve an item (Furl lacks this retrieval feature)

see more here (scroll to the heading Personal vs Social)

Even though there is no authority system the subject (tag) search still provides relevant results, the only thing is that the index is a bit messy (there are no rules)..tags are not solely used as subject terms, they are just labels in the end.

Sometimes when I search library databases for a specific subject term, it returns zero results, then I notice by chance (from another article) that I was using the wrong term according to their subject index. For example, instead of using the subject term “RSS”, they use “electronic publishing” or “communication technologies.” Now to my mind, in some instances, a user defined vocabularly maybe be more savvy on terminology compared to a traditional indexer of a controlled vocabularly database. I guess they may want to accumulate a certain amount of articles before they warrant applying a new subject term to their vocabularly…so there may be a time factor also.

So the question is “effectiveness.”

Would you use it in a business setting to replace your taxonomy?

At the moment, I’m not so sure, although a folksonomy is handy in the first part of building a taxonomy, ie. gathering user-defined subject terms to define a set of suitable terms for retrieval…I’ve touched on this before, more to add…

For example, in an experimental setting, tags can still be user-defined, but then a moderator comes along and controls the emergent vocabularly…sculpturing the vocabularly as it happens…then everytime you tag future pages you can choose one from the vocabularly and apply a suggested tag also which goes to the moderator (you could also apply suggested tags for older content as its context or specificity may have changed).
Maybe this could work in a business setting but it just wouldn’t scale in the WWW.

…see these posts for more:

Enterprise Distributed Categorisation: I Get on the Folksonomy Bandwagon
The Cognitive Cost of Classification
Using mapped folksonomy to break corporate silos
IBM’s Intranet and Folksonomy
Taxonomy vs. Folksonomy Debate Heats Up

Folksonomy structure

Folksonomies have a flat-file index (all terms are on the same one level). This doesn’t mean they can’t be cross referenceed as related in some way, but it does mean they are not ordered in a hierarchy or parent-child relationship - so I guess they are neither a taxonomy nor controlled vocabularly.

NOTE: to my understanding traditional databases with controlled vocabularies allow you to browse a subject field, this subject index is usually arranged alphabetically…sometimes you can search this index. A subject index can be organised in another way that is more useful than browsing an alphabetical list…it can be arranged in a hierarchy with synonym control where terms can be explicitly related, this is generally a thesaurus…they can also be arranged in other useful ways.

see more for taxonomies.

The flat-file index is made up of all the users “tags”…at the moment on del.icio.us you have to use alternate interfaces to search for tags (and see the content within)…see Technorati tags, Tag central, del.icio.us tag search.
Once you have found your tag and viewed the contents within, it does not allow you to further search within a particular tag.

Alternately you can use del.icio.us itself by typing in a potential tag term in the URL, such as “hats”, eg. http://del.icio.us/tag/hats.
Or you can browse the popular tags on the front page (also see trendalicious! and populicious)…or even jump from bookmark to bookmark looking at tags in other public accounts.

To search for text in a page (del.icio.us only searches the text in the URL and the extended description) there is a way to export content to gmail, very clever to take advantage of gmail’s strong search functionality.

Some social bookmark managers, like Furl, can offer searching within a tag, but only within a personal account, this can’t be viewed in the collective version of tag search engines such as Zniff (searches Spurl) and Furl (need a log-in).
These both search the full-text of human indexed pages, that is, webpages chosen by a collective social group - so it’s a cleaner version of the web (but not as exhaustive).
Problem is that they don’t search for tag labels, and when you find a tag you like, then to search within that section…although, as mentioned before, in Furl, the personal version allows you to search within a folder/tag.

When it comes to searching, it seems that what del.icio.us lacks, Furl and Zniff offers, and vice versa. see this post again…and also the functionality of the personal account, in Furl, is more sophisticated than the collective version.

For now…

Social bookmark managers are tools that don’t control the tag index (as that is counter to what a folksonomy is - an uncontrolled, user defined vocabularly) but contribute to the building of synonym control by the power of suggestion or recommendations.
From this post on the offerings of the del.icio.us posting bookmarklet:

“…a list of popular tags for the page you’re bookmarking, your full tag list, and some popular tags in general.”

Another way to help synonym control is that once you have bookmarked your page, you can click on the “…? other people” link to see what others have linked…but this is after the fact and therefore not reliable.
A great tool I’ve mentioned before is del.icio.us linkbacks…this is before the fact, allowing (via a bookmarklet) to see who has bookmarked the page you are on and the tags applied.
You can extend this by using Durl to get an RSS feed of new people that bookmark a page to see the tags they apply.

Another tool, not so much for synonym control but for mistakes is Tagstemmer, it cleans up similar spelt terms.

Lastly another tool is gre.gario.us…it is used to find similar content to your del.icio.us account (other accounts have to have at least 2 of the same tags), making it a great social tool. But it is also a tool to see what tags other people are using, so you can inturn refine the consistency of tags, ultimately augmenting the public vocabularly that is the folksonomy.

Personal OPAC, personal journal databases, personal anything…

Furl and Spurl seem to be the only bookmark managers that are closer to a traditional library catalogue or a journal database…although I think del.icio.us has a more social intuitive feel. Saying this I have only had limited use of other popular sites such Connotea or CiteULike (seem more akin to journal databases)

The reason I say this is that the folders in Furl can be used to mimic a broad version of the DDC classification system and the keyword field can be used as subject terms.

NOTE: are people aware of this field when they bookmark in Furl…you can search Furl by prefix with an assigned term, eg. Keyword: Hats…as long as you have apllied a term/s in the keyword field at the time of bookmarking.

  • folders/tags assimilate browsing the shelves, or DDC field
  • keyword prefix assimilates searching in the subject field
  • leaving the keyword prefix out of the search is like searching the free full-text
  • you can also search by keyword prefix or free text within a folder/tag
  • although it doesn’t allow searching within two or more folders at the same time

This arrangement could also be done in del.icio.us, you’d just have to distinguish between the classifications set of tags from the subject term set of tags…maybe using a capital letters for the former and lower case for the latter.

Something Furl lacks in it’s social features is searching everyone’s folders (tags) or everyone’s keyword (prefix) index…well the keyword prefix isn’t a social feature full stop (you can’t share keywords as they don’t have social tags). I’ve mentioned before (scroll to the heading Personal keyword index anyone?), that it would be great to view a list of your keyword index.

But is this type of thinking trying to stretch a one-level flat file index to its limits?…it’s the kind of thinking that leads us back to sub-folders…or can you simply just have 2 types of tags?

Another great feature, in Furl and some of the others, is that you can apply comments or clippings to each bookmark, this would be similar to adding a table of contents to bib records in an OPAC or adding an abstract in a journal database…del.irio.us, a clone of del.icio.us, also incorporates this feature.

Ideal system

  • Search or browse for tag labels (in the alphabetical subject index) and view the content within
  • also offering related tags
  • adding multiple tags together
  • refine search within a tag or a set of tags
  • search full-text
  • results returned in tag clusters
  • refine search within a tag/s

Apply all of the above to a meta-Social bookmark search tool…returning results within multiple social bookmark managers.

Extensible system

Personalising a web-site, database or OPAC by tagging the content…build your own personal version adding to the aggregated version.

Include aggregated version as an alternate subject term search field or as an alternate navigation browsing tool…see here for more

see these posts for more:

library websites and folksonomies
Grassroots Cooperative Categorization Of Digital Content Assets: Folksonomies, What They Are, Why They Work
Folksonomies, internationalisation and libraries
Del.icio.us really is delicious
Folksonomies
MLS February Tech Summit on Social Bookmark Services
Librarians need to be flexible

all consuming is an alternate system to searching/browsing amazon.com…users bookmark webpages from amazon.com and apply tags for retrieval. This is only an alternative to using amazon.com for searching/browsing…it could also be extensible by being integrated into the amazon.com website itself…hopefully this is where folksonomies will lead…

Innovative ideas

Consumerpedia, a wiki based web-site where users create a hierarchial folksonomy to categorise the content for retrieval…here are two reviews…here is a blog post from Consumerpedia and their help page.

Another application in building a communal website and using user defined tags for navigation/retrieval is Tagwhere…it seems you have 2 levels of location tags and a subject label tag…you can also ping from a blog post to Tagwhere…see their about page.

Others are 43 things and grat.uito.us…the content in both these social bookmarking sites are specific to wishlists.

Some of my related posts:

User tags for the: WWW and OPAC
Labelling and arranging for future reference: control…
Folksonomy before Taxonomy
Folksonomies in contrast..
Tagging within a group environment
Visually emerging vocabularies…
Folksomonies with rules..
Some thoughts on Furl and del.icio.us…(and folksonomies)
More on del.icio.us and Furl…
Date with del.icio.us!
Zniff clusters

10 Comments »

The URI to TrackBack this entry is: http://libraryclips.blogsome.com/2005/04/15/social-bookmarks-so-far/trackback/

  1. Good points and well taken. Our Zniff engine is obviously in its very early days, but we will be offering more powerful ways to “dig in” as things progress.

    To see where we stand on the tags / folksonomies issue, you might want to check out this blog post I posted earlier this week on the subject:
    Coming to terms with tags: folksonomies, tagging systems and human information

    Comment by Hjalmar Gislason — April 15, 2005 @ 9:18 am

  2. […] it my radar until this morning, but it quickly went into my Bloglines list. Today’s posting on social bookmarking and folksonomies offers a fairly extensive rou […]

    Pingback by blogdriverswaltz.com » Folksonomies Overview — April 15, 2005 @ 3:14 pm

  3. The reason for this imperfection in indexing is that many people are doing the tagging (everyone has different ideas in judging the aboutness of something)

    Yes, and what’s more, someone may not even tag “aboutness” of the tagged item, but some subjective characteristics like “toread”, “inspirational”, etc. Tagging is something broader than just classification into multiple categories.

    Comment by quirkyalone — April 16, 2005 @ 11:11 pm

  4. […] ears ago, but a month later it is still very much worth mentioning. Library Clips wrote a meta-post on social bookmarking back in April that blew my mind with its com […]

    Pingback by Information Wants To Be Free » What’s happening in social bookmarking — May 5, 2005 @ 12:06 am

  5. links for 2005-05-08

    tagSense.net (tags: del.icio.us) Om Malik%u2019s Broadband Blog � Internet Anxiety Disorder Anyone? funnen via rymdimperiet. (tags: internet att_läsa) Contentious…

    Trackback by Malins blog — May 8, 2005 @ 11:19 pm

  6. […] is scalable and usable(effective) - in essence a user defined subject index for the web. Library clips :: Social bookmarks so far… :: April :: 2005 […]

    Pingback by HYPERGURU » Library clips :: Social bookmarks so far… :: April :: 2005 — May 20, 2005 @ 11:40 pm

  7. Very good!

    Comment by flower — June 20, 2005 @ 2:00 am

  8. Time ago I thought… Would be great If I have access to a social bookmarking search engine, where all links have a relevance similar to the “pagerank” algorithm, but based on people interests.

    I think that someone else already put my idea on rails … :(

    This is the site: www.thagoo.com

    Larry.

    Comment by Larry Green — April 27, 2007 @ 1:52 pm

  9. http://socialposter.com is a service that lets you post a web page to numerous social bookmarking sites.

    Comment by Dan Griffin — April 30, 2007 @ 2:28 am

  10. thought tagging & social bookmarking are also good way to give in contribution for websites.

    http://www.resumepromo.com

    Comment by benz — July 23, 2007 @ 8:44 am

RSS feed for comments on this post.

Leave a comment

Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>



Anti-spam measure: please retype the above text into the box provided.

Please note that comments are moderated and will                 not therefore appear immediately.
                Please do not repost.


Library clips
Library clips Subscribe by Email                                                      

Get free blog up and running in minutes with Blogsome | Theme designs available here