Collective intelligence Assignment #2 – Can social cataloguing aid library metadata creation?

By Nicole Basaraba

Social tagging, or folksonomy (a portmanteau of folk and taxonomy) coined by Thomas Vander Wal is a form of participatory literacy. It gained popularity in 2003 with the social bookmarking site “del.icio.is” (Lu, Park & Hu, 2010) and two of the largest websites employing social tagging and focusing on books are LibraryThing and Goodreads. Social tagging is a process where users add metadata often termed tags or keywords to content.

The three texts (book marked on my delicious account) I chose to examine are: 

1. a YouTube video which shows how social tagging works in LibraryThing,
2. a journal article that explores whether social tagging in LibraryThing is comparable or useful for expert-created subject headings for the Library of Congress, and
3. a blog that outlines why Goodreads – which relies more on non-textual metadata – rather than LibraryThing is the largest social book sharing website.  

These texts showed that social cataloguing, while uncontrolled, can help: Internet users personalize their workflow, contribute new terms to standardized library taxonomies, and help people form communities online.

Video: “What is social cataloguing?” – Tim Spalding

In this video Tim Spalding mocks his own ethos. He has Classics degree, but describes his creation of LibraryThing as “Library Science practiced without a degree”.

Spalding started LibraryThing in 2005 and now it is a company based in Portland, Maine with over 1.7 million users and 88 million books catalogued. It adds books from 690 libraries around the world and from Amazon.

Spalding defines social cataloguing as a situation where “personal cataloguing goes social”. His definition is vague, but he explains that LibraryThing’s main purpose is for personal cataloguing where users can add any tags they think describe the book’s topic and content. It displays information in an ‘Excel sheet’ type format with detailed information including book author, title, ISBN, book awards, and other categories. Spalding emphasizes that LibraryThing does not suggest tags so that users organically generate all the tags. With this system, users can search for books and view and interact with someone else’s library allowing members to grow into a community.

In addition to the community aspect, Spalding notes that the personal cataloguing reveals a lot about a person. The tag cloud shows users’ topical interests and the statistical metadata provides users with insight about themselves.

Journal Article: “User Tags versus Subject Headings: Can User-Supplied Data Improve Subject Access to Library Collections?” - Caimei Lu, Jung-ran Park, and Xiaohua Hu, Drexel University, USA

Lu, Park and Hu (2010) conducted a study comparing social tags created by users from the LibraryThing website with the Library of Congress subject headings determined by expert cataloguers. Since social cataloguing is unregulated, researchers are concerned with the quality of the tags.

Figure 1. Benefits and limitations of expert-created compared to socially-created metadata.

	Expert created metadata (Library of Congress subject headings)	Social cataloguing (LibraryThing tags)
Benefits	High quality Limited scope	Time efficient More consensus on subject scope Adaptable to new vocabulary
Limitations	Costly to produce	Uncontrolled/limitless

Experts apply controlled vocabularies to metadata to summarize resources after reading the books so this process is time consuming and costly. The language used for Library of Congress subject headings (LCSHs) is limited because it is often highly technical and specialized, which requires users to have knowledge on these terms in order to search for books (Lu, Park & Hu, 2010, p. 764).  

Social tagging doesn’t involve any metadata standards and users apply their own descriptors. Since this threshold is lowered, a large number of resources can be tagged in a short amount of time (Lu, Park & Hu, 2010, p. 764). Social tags can include new popularized vocabulary, non-dictionaries words, and words that have meaning only to the individual user.

For example, in del.icio.us we used the tags “NMN” and “For:JessL” for this assignment so that we could find each other’s content and links. However, these tags don’t really mean anything to anyone else who hasn’t taken the course before. A study by Kipp and Campbell found that more than 16% of del.icio.us tags were non-subject terms, which suggests that users want to attach personal resource management information to documents or books to make searching for them easier (Lu, Park & Hu, 2010, p. 772).

The key finding in Lu, Park and Hu’s study (2010) between the two cataloguing systems is that “overlapping terms only comprise a small portion (about 2.2%) of the entire tag vocabulary. That is, about 97.8% of [user-generated] tags cannot be found in LCSHs for the same collection of books” (p. 770). This demonstrates that uncontrolled tagging results in limitless categories.

Lu, Park and Hu (2010) came to three main conclusions:

• “user-generated tags may help bridge the gap between professional and public discourse by providing a source of terms not included in authorized terminologies and controlled vocabularies […]
• social taggers might help enhance the subject access to library collections by describing library resources with terms different than those used by experts […]
• users view classification as a holistic process closely tied to themselves and their work” (p. 776-777).

In other words, although social tagging is uncontrolled, it can help: contribute to expert-created metadata, improve online search ability for books, and add an element of personalization for individual users’ work. 

Blog: “Social cataloguing – a discussion on LibraryThing and Goodreads” – Simon Barron, Librarian

LibraryThing and Goodreads are the two most popular social cataloguing websites. Goodreads surpasses LibraryThing with approximately 17 million users and 550 million books (Barron, 2013), which raises the question as to whether Goodreads is superior to LibraryThing.

Simon Barron notes that Goodreads focuses more on the social networking aspect because it has a more visual interface than LibraryThing’s text-based user interface. Barron’s description of the two networks can be understood from a phenomenological perspective. Barron notes that design may be a major factor in Goodreads’ popularity:

“Goodreads tends more towards the use of terminology related to cataloguing print materials and to old paradigms of organisation: 'shelves', 'friends', 'community' are familiar and comforting words associated with cosy, physical libraries and lovely, welcoming bookshops. LibraryThing, by contrast, uses words that actually apply to digital materials and which, while more technically accurate, are less welcoming: 'collections', 'members', 'contacts', etc.” (para. 10).

While Goodreads is more popular, LibraryThing’s metadata provides more intelligent book recommendations for each user based on the books in their library rather than Goodreads, which relies on the user attributing star ratings (1-5).

Barron concludes that social cataloguing: allows users to organize books quickly and easily, removes the reliance on trained cataloguers, and makes cataloguing more subjective, democratized and social.

Conclusions

LibraryThing is advantageous for users because it: gives them a personalized digital library where they can easily search for books according to their own semantics, allows them to learn more about books they’ve read, and provides greater insight into their own interests. LibraryThing’s focus on organic user-developed tags creates categories or new taxonomies that could potentially be used by libraries to enhance search ability for books. For example, Spalding (in the second video) gives is the fiction genre “cyberpunk” – a category not recognized by the Library of Congress, which instead uses the “Science Fiction”.

Goodreads surpasses LibraryThing in terms of number of users and books because the website design allows it’s users to engage in more phenomenologically familiar and simplified book sharing and reviewing practices. However, LibraryThing’s social cataloguing is superior because the metadata provides better book recommendations and more detailed information on titles.

There is potential for libraries to incorporate social tagging into their cataloguing to improve search ability and user interaction. For example, the University of Pennsylvania created ‘Penntags’, a locally developed software for social bookmarking; the Ann Arbor District Library developed the ‘Social Online Public Access Catalog’; and Danbury Library was the first to incorporate LibraryThing widgets into its catalogue (Lu, Park & Hu, 2010).

New media technologies allow users to create their own ‘narratives’ as they can assign personalized tags and meaning to works they’ve read. They can do this quickly to a large volume of books and it allows them to improve personal search efficiencies. This online participation (literary smart mob) democratizes book taxonomy into a folksonomy and creates online communities and shared knowledge, which contributes to everyone’s digital literacies.

References

Barron, S. (2013, August 6). Social cataloguing: a discussion on LibraryThing and Goodreads. Undaimonia. Retrieved from http://undaimonia.blogspot.ca/2013/08/social-cataloguing-discussion-on.html

LibraryThing. Retrived from https://www.librarything.com

Lu, C., Park, J., and Hu, X. (2010). User tags versus expert-assigned subject terms: A comparison of LibraryThing tags and Library of Congress Subject Headings. Journal of Information Science, 36 (p. 763-779).

Spalding, T (Speaker). (2009, November 30). LibraryThingTim [Episode 1/6]. What is Social Cataloging? Retrieved from http://www.youtube.com/watch?v=kCU_UkUKZI8

Spalding, T (Speaker). (2009, December 1). LibraryThingTim [Episode 2/6]. What is Social Cataloging?. Retrieved from http://www.youtube.com/watch?v=Kt94e9mX320