Course:LIBR557/2020WT2/folksonomy

Folksonomy

Folksonomy is a knowledge organization system that comprises a collection of user-generated tags applied to digital information (websites, images, text, etc.) for easy identification and retrieval.

The practice of applying tags is known as social tagging or collaborative tagging (Yu & Chen, 2020). Tags reflect the user’s conception of an item and allow the user to apply descriptive terms that are meaningful to them (Porter, 2011). The collections and connections that result from this practice form a folksonomy (Porter, 2011; Holstrom, 2018).

History

A portmanteau of the terms “folk” and “taxonomy”, the term folksonomy was coined by Thomas Vander Wal in 2004 (Holstrom, 2018; Besseny, 2020).

Folksonomies first appeared in the mid 2000s on digital bookmarking sites such as del.ici.ous, and image repository sites such as Flickr, as users applied descriptive tags to online items for personal reference, organization, and retrieval (Yu & Chen, 2020). Folksonomy quickly became a responsive way of organizing huge amounts of rapidly growing and changing information (Porter, 2011).

Folksonomy as a knowledge organization system

Folksonomy is a flat knowledge organization system and does not follow a hierarchical structure. There are no established or controlled parent/child relationships or synonymous relationships between terms (Bullard, 2018). The tags that make up a folksonomy are not created by professional indexers or organized according to an authority control or controlled vocabulary and are free-form and limited in their semantic relationships with each other (Yu & Chen, 2020).  

Folksonomies are instead made up of tags which reflect the language of its users and so a user-generated vocabulary emerges (Yu & Chen, 2020). These tags generally follow a ‘power law’, meaning that there is a concentrated, popular use of a few terms and then a ‘long tail’ of terms that are used much less frequently (Holstrom, 2018). In this way, despite the diversity of individual users and the tags they apply to items, researchers have found that the distribution of tags reaches a proportional stability after a certain number of tags are assigned to an object (Golder & Huberman, 2006)—this is especially the case in broad folksonomies in which many users are tagging the same items (Holstrom, 2018).¹

Querying folksonomies

Searching

Searching folksonomies often takes the form of browsing rather than a search for specific terms and phrases (Trant, 2009) and knowledge discovery is often serendipitous (Porter, 2011). This browsing can be done by searching within categories of tags, single tags, or by searching the tags posted by a specific user (Trant, 2009). Searching tags has been compared to keyword searches in that users search for items using descriptive terms (Golder & Huberman, 2005), often using search mechanisms similar to search-engines (Navarro Bullock, et al., 2018). Folksonomies are not often searchable using Boolean queries (Yi & Chan, 2018).

Results 

Folksonomy search results are often ranked based on frequency and popularity of tags. These can be displayed visually in the form of tag clouds with the use of different font sizes to denote the ranking of results (bigger fonts denote more popular tags) (Besseny, 2018). Other methods of ranking results in a folksonomy are based on algorithms which consider networks of hyperlinks and user behaviours (Navarro Bullock, et al., 2018).

Precision and recall

Precision and recall are negatively affected by the user-centric focus of folksonomies. Aspects such as grammatical differences, misspellings, and conflation of homonyms, etc. decrease the precision of results (Bullard, 2018) while the prevalence of popular and personal tags result in inflated recall (Macgregor & McCulloch, 2006 cited in Porter, 2011).  

Strengths and weaknesses as a knowledge organization system

Both the strengths and weaknesses of folksonomy are derived from its user-focused nature.

One of the strengths of folksonomy is that it is able to keep up with the fast pace of collection growth of content-sharing platforms and to organize that content quickly and with low effort by a wide variety of users (Bullard, 2018) due in part to the act of tagging being a simple one (Yu & Chen, 2020). The tags that make up a folksonomy represent the diversity of the users (Bullard, 2018), are inclusive (Gordon-Murnane, 2006 cited in Porter, 2011) and represent the user’s language (Holstrom, 2018). Because of the expertise and interest of the users tagging items, retrieval success is generally increased beyond that of full-text searching (Bullard, 2018).

The challenges of folksonomy centre around the overall inconsistency of a vocabulary created by many different users (Trant, 2009). User diversity in approaches to spelling, grammar, specificity, and language lead to the creation of tags that are separate from each other, but which mean the same things, affecting the retrieval of search results (Bullard, 2018). These aspects result in poor precision and recall (Bullard, 2018). Also impacting precision and recall is the emphasis on searching single-word tags (Smith, 2004 cited in Trant, 2009). This can be seen in Rosenfeld’s 2005 example of searching Flickr with the single tag “summer” which retrieves thousands of unranked results (cited in Yi & Chan, 2009).

Current research and future developments

Researchers have attempted to address some of the weaknesses of folksonomy and of controlled-vocabulary classification systems by matching free-form tags with professionally indexed subject headings such as the Library of Congress Subject Headings to increase knowledge discovery (Yi & Chan. 2009). Researchers have further proposed the development of automatic methods which would map tags to subject headings (Yu & Chen, 2020). Research and practical examples have explored the “curated folksonomy” which is the application of human judgement to folksonomy to give it structure and to solve challenges such as synonymity (Bullard, 2018).

Tagging systems have already been integrated within discovery layers of library OPACs (Porter, 2011) and are integral parts of the social aspects of library discovery layers such as BiblioCommons.

Bibliography

Besseny, A. (2020). Lost in Spotify: Folksonomy and wayfinding functions in Spotify’s interface and companion apps. Popular Communication, 18(1), 1-17. DOI 10.1080/15405702.2019.1701674

Bullard, J. (2018). Curated folksonomies: Three implementations of structure through human judgment. Knowledge Organization, 45(8), 643-652. DOI: 10.5771/0943-7444-2018-8-643

Golder, S., & Huberman, B.A. (2006). Usage patterns of collaborative tagging systems. Journal of Information Science, 32(2), 198-208. DOI: 10.1177/0165551506062337

Holstrom, C. (2018). Social tagging: Organic and retroactive folksonomies. In Proceedings of the 18th ACM/IEEE on Joint Conference on Digital Libraries (JCDL '18). Association for Computing Machinery, New York, NY, USA, 179–182. DOI:https://doi-org.ezproxy.library.ubc.ca/10.1145/3197026.3197065

Navarro Bullock, B., Hotho, A., & Stumme, G. (2018). Accessing information with tags: Search and ranking. In: Brusilovsky P., He D. (Eds.), Social information access. Lecture notes in computer science (pp. 310-343). Springer. https://doi-org.ezproxy.library.ubc.ca/10.1007/978-3-319-90092-6_9

Porter, J. (2011). Folksonomies in the library: Their impact on user experience, and their implications for the work of librarians. The Australian Library Journal, 60(3), 248-255. DOI: 10.1080/00049670.2011.10722621

Trant, J. (2009). Studying social tagging and folksonomy: A review and framework. Journal of Digital Information, 10(1). Retrieved February 20, 2020, from: https://journals.tdl.org/jodi/index.php/jodi/arti cle/view/269/278

Vander Wal, T. (2005). Folksonomy Definition and Wikipedia. Off the Top. Retrieved February 20, 2020 from http://www.vanderwal.net/random/entrysel.php?blog=1750

Yi, K., & Chan, L.M. (2009). Linking folksonomy to Library of Congress Subject Headings: An exploratory study. Journal of Documentation, 65(6), 872–900. doi: 10.1108/00220410910998906

Yu, W., & Chen, J. (2020). Enriching the library subject headings with folksonomy. The Electronic Library, 38(2), 297-315. https://doi-org.ezproxy.library.ubc.ca/10.1108/EL-07-2019-0156

Footnotes

1. Folksonomy can be broad (when many different users tag an object) or narrow (when only the creator does so) (Trant, 2009).