You are on page 1of 20

Running Head: CONTROLLED

VOCABULARIES AND FOLKSONOMIES

Controlled Vocabularies and Folksonomies: Broad Connections and Implications for the Future Emporia State University LI 804XO May 4, 2012 Professor Ann O'Neill

Controlled Vocabularies and Folksonomies:

CONTROLLED VOCABULARIES AND FOLKSONOMIES Broad Connections and Implications for the Future Introduction

Around 2005 a new phrase was penned by Vander Wal (2007) called folksonomy. Folksonomy is a combination of the words folk and taxonomy and is the result of personal free tagging of information and objects for one's own retrieval (Vander Wal, 2007). Since then, much research and debate has been done concerning the potential folksonomies have within the library and information science communities. This paper aims to give a brief introduction to folksonomies, their advantages and disadvantages in relation to traditional controlled vocabulary use, and, most importantly, to discuss whether their adoption within Library and Information Science will lead to greater information seeking and access success. What are Folksonomies? It has always been the case that creating controlled vocabularies is a limiting, subjective process that is open to bias. By definition, controlling a vocabulary is controlling the words with which people use within it, thereby reducing the number of possible interpretations (Porter, 2005). No matter who the cataloger is, their background experiences will influence knowledge and worldviews. These factors all have influence on how an object is described. The success of the information seeker then will depend heavily on their ability to relate to and navigate within the original catalogers framework. One person's view or choice in how an item is described can then hinder retrieval by others. The idea that users of information might also contribute to the description of that information is, then, intriguing. Since the mid 90s researchers have been thinking about how users descriptions of data might be aggregated to form an overall view of the data. This overall

CONTROLLED VOCABULARIES AND FOLKSONOMIES

aggregated view, it is important to note, does not form a consensus view, but simply a broader view (Speller, 2007). The work of Mathes (2004) gives us perhaps the best introduction to tagging and what folksonomies are. Mathes (2004) explains that there are three different agents possible in the creation of meta-data: professionals (catalogers), authors, and users. In the traditional form of meta-data creation, catalogers were the prime creators. With the invention of the internet, users are now doing their own cataloging. User tags are non-hierarchical, free-form, keywords or terms assigned to describe a piece of information (Wikipedia contributors, 2012). Users tag meta-data for many reasons, both within and outside the sphere of libraries. When multiple users tags are aggregated they form a folksonomy. Folksonomies can then be used in two ways. The tag creators can use tags to search and retrieve their own information at a later date. Or other people can use these tags for information searching and discovery. Folksonomies on the Internet There are now many implementations of folksonomies on the internet. Two of the first, and most famous, examples are that of Del.icio.us and Flickr. Del.icio.us is a social bookmarking website that allows users to tag and share websites they bookmark. By using Del.icio.us users share what websites they like the most. The more times a website is bookmarked the more likely it is to be seen by another user on Del.icio.us and thus be bookmarked again. In addition, tagging allows users to search for bookmarks based upon other users descriptions. Flickr is a photo-sharing website. Similar to Del.icio.us, Flickr allows users to tag and

CONTROLLED VOCABULARIES AND FOLKSONOMIES

describe pictures. This information can then be shared with the online community. Tags, titles, geo-location, and other aspects of pictures can then be searched. Other websites on the internet are using tagging in varying ways. Websites like Bibsonomy are using user tagging to share scholarly works available on the internet. Websites like YouTube and Last.fm use tagging to connect people with music and videos. Controlled Vocabulary on the Internet It is important to distinguish folksonomies, user generated vocabularies, and distributed classification from a number of other ways in which websites are using description and sharing of information. Many websites, like Amazon.com, now use controlled vocabularies along with faceted searching to classify products. This is an example of a taxonomy. While Amazon does allow user input in tagging and rating items, their primary search results are based upon an internal classification and search scheme that uses categories and subject headings. Open Source and Crowd-sourcing Other websites are doing similar things to folksonomies through crowd-sourcing and open source. Instead of leaving it to the professionals, companies and organizations often allow people to take part, or volunteer, in the creation of something. In the case of open source there are many contributors and many beneficiaries. In the case of crowd-sourcing, there are many contributors but few beneficiaries. Perhaps the largest example of an open source project is the Linux operating system. This operating system, in contrast to proprietary developments like Microsoft Windows, is developed primarily by volunteers that then share their developments with others. Through this

CONTROLLED VOCABULARIES AND FOLKSONOMIES

collaborative process a computer operating system is developed and improved. Many people can and do contribute. Many people can and do use the resulting product. In contrast to open source, crowd-sourcing is the use of volunteers in developing a product that will benefit a company or organization. One major example of this type of outsourcing is the SETI@home project. It uses volunteers computers processing power, over an internet connection, to help researchers search for extraterrestrial life. The volunteer is gaining no direct benefit by volunteering their computer. But there is often times an indirect benefit or interest that gives the volunteer reason to help. While both open source and crowd-sourcing are related to folksonomies in that they are helping to share information, folksonomies use distributed classification. The distinction is that folksonomies are related to the search and open source and crowd-sourcing is related to creation. Advantages Consensus from the Bottom Up There are several advantages that folksonomies have over controlled vocabularies. Perhaps the greatest is (Shirky, 2005) that in a folksonomy consensus views of the world are reached in a bottom-up approach. Since the index terms are created by the users then it would follow that other users will be able to relate more to such terms and subsequently find what they need. A top down approach will always result in a disconnect between where people start their search, at the bottom, and where catalogers, at the top, expect the search to go. Almost Faceted Classification (Speller, 2007) The nature of folksonomies is non-hierarchical and not subject to taxonomic classification. Therefore, it is easy for folksonomies to identify an item as being about multiple

CONTROLLED VOCABULARIES AND FOLKSONOMIES

things simultaneously. The current ability to use almost limitless subject headings to describe an item is similar. But, unfortunately, no current tagging systems create true faceted systems that allow for multiple navigable hierarchies. Instead, there is no hierarchy. Everyone can Take Part Since folksonomies are primarily online, there is no geographic limitations to who can take part in a folksonomic effort. Leaving aside issues of income and access, anyone in the world can conceivably create a Del.icio.us account and start tagging. This can be seen as an advantage over traditional classification schemes, like the Dewey Decimal System, that are imposed upon cultures different from where the system originated. There are still fundamental problems related to linguistics in a folksonomy. Tagging is primarily done in english and many systems that use folksonomies cannot account for non-Latin alphabets. But at least in a folksonomic system all tags are connected to the user that created them. Shirky (2005) argues that at least in a distributed classification system the people creating meta-data tend to be visible members of the system. This allows their possible biases to be understood and taken into account. Anyone Can Easily Take Part Mathes (2004) argues that folksonomies work chiefly because they have a low barrier to entry. Tagging has a lack of guidelines and rules that makes entry almost effortless. In comparison, the latest addition of the Anglo-American Cataloging Rules is 750 pages long (AACR, 2005). This ease of access opens cataloging up to two related segments of society. First, the ease of access makes the casual cataloger a participant. These taggers broaden the field of participants. Second, the ease of access opens up cataloging to those with little or no education. While this open access can create problems through the diversity of taggers backgrounds, no one

CONTROLLED VOCABULARIES AND FOLKSONOMIES can argue that diversity in consensus making is a bad thing. Keeping up With the Changing Vocabularies

Many have cited the advantage folksonomies have in their flexibility and ability to reflect changing terminologies. Folksonomies are not subject to the process in which new terms are selected and added to controlled vocabularies. New terms can be tagged just as soon as they are thought up. There is no vetting process or efforts needed to fit the new term within a taxonomy. This ability is especially helpful in the rapidly evolving online environments tagging is used in. Serendipity Because of the flat structure of tagging, related tags are easier to see. A thesaurus or subject headings can allow for a certain degree of serendipity within a search. But tags allow for a much more broad view of related topics. While this does not necessarily help users find specific topics, it can help in the sense-making process of self discovery. Cost With thousands of volunteers doing the categorizing for you costs are reduced significantly. It would be wrong to call tagging entirely free. Websites have server costs. But with folksonomies there is little cost due to two factors. There are no longer any rules to learn and apply for the creator of tags and all tagged meta-data is freely shared by all. Disadvantages The Problems with Coming to Consensus The advantages of coming to a consensus viewpoint in folksonomy terms can be easily limited by several factors. There are conditions to reaching consensus. First, a diversity of opinions needs to be represented. If all possible opinions are not given, then the consensus

CONTROLLED VOCABULARIES AND FOLKSONOMIES

reached could be missing the one term that would have been the best. Second and third, independent decision making, coupled with decentralization of power is needed. If people form their opinions based off of the prompting of others then peoples opinions are not truly being represented. Finally, a way of aggregating opinions needs to be available (Speller, 2007). If there is no system in place for people to count the vote then a voting result cannot be reached. Who is doing the voting? Is it really consensus? Another connected problem with folksonomies is found when we start to look at who is doing the voting. Some researchers of folksonomies argue that only a select group of people get to vote. Just who are the people that are creating these folksonomies? Based off of the website demographics and time it takes to work on folksonomies, they are well connected, technologically savvy, opinionated and young. If we consider who might not be participating in the tagging process we then must consider whether a true consensus is being reached. Consensus is based off of the consent of all to come to a decision. Not all parties need agree for consensus to be achieved. But if not all parties are participating, then consent, and therefore consensus, cannot be given. Still, this flawed consensus of some people is perhaps better than giving over that power to a few catalogers. Keeping the chaos out Even a controlled vocabulary will have problems with consistency and accuracy across different catalogers. But when a large group of untrained volunteers are doing the indexing, problems of inconsistency and inaccuracy can become unwieldy. Because of a lack of control inherent in a folksonomy there are several factors having to do with overlap that it does not address. One and all can create disorder in terminology that will actually be a hindrance to

CONTROLLED VOCABULARIES AND FOLKSONOMIES people finding the information they seek. Synonyms and homonyms

Synonyms and homonyms are two such factors. A synonym is a word that has the same or similar meaning. Unless two pictures on Flickr are tagged as both feline and cat then those searching for one search term or another will find only one or the other result. Homonyms are words that are spelled the same but have different meanings. Tagging does not allow a user to distinguish between a bow and arrow and a bow to honor someone. In reality, these types of problems may not be as large as critics of folksonomies believe. One study (Spiteri, 2007) of three folksonomic sites found that the actual proportion of homographs and ambiguous tags each constitutes fewer than one-quarter of the tags. While certainly a problem to address, it does not necessarily break the system. The Longevity of Tags Because folksonomies can introduce new terminology quickly, they can also be subject to an ever shifting tag cloud in which the user must know the passphrase to get in. Unless new terminology tags are matched up with old previously tagged items, old tags will not match up with new ones and the user will only get partial results. Capitalization, Plural vs. Singular, Misspellings, and multiple word Tags Even when people choose the same basic terminology and meanings behind tags, variations in how those same tags are typed can become a problem. Taggers do not always use capitalization, they do not always properly spell the term they are thinking of, and they do not always choose a singular or plural form of the term. In addition, sometimes there is a need to tag something using multiple words for a single

CONTROLLED VOCABULARIES AND FOLKSONOMIES

10

tag. Some users will use spaces between the words. But some folksonomic websites do not allow for spaces. A new space will indicate a new tag. As a result, some users will use different punctuation marks to indicate a space. Others will simply run words together, as in thisisacat. Personal vs. Social Tagging Tagging works because it strikes a balance between the individual and social (Sinha, 2006). But tagging is not always done with the benefit of all in mind. There seems to be two types of tagging that are done. Some taggers tag things to help themselves better organize and catalog their own libraries of information. While others wish to share the information they have gathered. There is overlap in these two types of tagging. There also, is not necessarily a conflict between personal interests and the interests of the whole. But a tag that is toread is not necessarily helpful to all in describing an object. In addition, connected to the issue of consensus making, are social issues of imitation and trendsetting (Sinha, 2006). When people tag they do it in a social setting in which others opinions count. Because of these social connections tagging is not done in a vacuum. People end up choosing tags to imitate those that came before them and whatever bias the original trendsetters had is compounded. Absolutism vs. Relativism Peterson (2006) argues that folksonomies reliance on relativism means that it is inherently flawed. Peterson argues that traditional classification schemes are based off of Aristotelian logic that A is not B. Peterson writes that using controlled vocabulary, Book A might be related to Book B, but a choice has to be made between them when classifying them. Peterson writes that while the act of one professional making a choice does introduce a bias of

CONTROLLED VOCABULARIES AND FOLKSONOMIES

11

viewpoint into the catalog, to include all viewpoints opens up a classification scheme to the inconsistency that allows a work to be both about A and not about A (Peterson, 2006). Golder & Huberman (2006) also talk about the sense-making that occurs when tagging occurs. They write, when one interacts with the outside world, one makes sense of the things one encounters by categorizing them and ascribing meaning to them. However, in practice, categories are often not well defined and their boundaries exhibit vagueness (201). In contrast, Porter (2005) argues that relativism is beneficial to information seekers because it includes the long tail of opinions. He argues that, without including everyones words it is unlikely that a system with ever-increasing information can meet everyones needs. He argues that controlled vocabularies marginalize many opinions, works, and ideas...by making them extremely hard to discover, or even worse, by excluding them. Meanwhile, Merholz (2004), argues for a more middle ground. He celebrates the userfocus of tagging while using the metaphor of desire lines in landscaping to discuss how relativism issues can be addressed. Desire lines are lines that people walk on the most, rather than where the landscaper wants people to go. Merholz writes, Ethnoclassification systems can similarly emerge. Once you have a preliminary system in place, you can use the most common tags to develop a controlled vocabulary that truly speaks the users language. Use the tags to understand how people consider the content at hand. Then you can pave the best paths to ensure findability. (Merholz, 2004) Shirky (2005) perhaps sums up all of these thoughts best as he writes, If...you believe that we make sense of the world, if we are, from a bunch of different points of view, applying some kind of sense to the world, then you don't privilege one top level of sense-making over the other. Instead you try to find ways in which individual sense-making can be helpful in an

CONTROLLED VOCABULARIES AND FOLKSONOMIES

12

aggregate. Shirky writes that by making an unnecessary choice in categorization we get, what he calls, signal loss. How can Folksonomies be improved? Controlled Vocabularies, like the Library of Congress Subject Heading System have had over 100 years to develop and improve. Some would say that they, sadly, have not improved that much in those intervening years. But the collaborative approach that folksonomies have seems to lend them well to the potential for vast improvements. Folksonomists will need to pay particular attention to any attempts to improve the systems in place. The overall impression is that distributed classification by many, coupled with guidelines of proper use of terms, will lead to more consistent, organized, and more useful folksonomies. Guy and Tonkin (2006) offer methods for improving folksonomies but warn that doing so may discourage users so they simply do not bother to tag further resources. Guidelines must be put in place that help to organize without restricting a diversity of opinion. This balance between diversity and rules may not always be able to be met. One current issue within controlled vocabularies and folksonomies is the degree of detail people use in tagging. One person may tag something as a bird while a birdwatcher may tag it as a blackcapped chickadee. These differences could mean that the two different taggers will find each others tags almost useless when searching. Hierarchical and structured vocabularies have made attempts, such as Sears Subject Headings, to provide for these differences in specificity. But ultimately, if one term, according to a set of rules, must be chosen, someone is going to have a harder time finding things. This is why it will be imperative that as folksonomies develop, they maintain diversity as hierarchies are created.

CONTROLLED VOCABULARIES AND FOLKSONOMIES Do Folksonomies serve the (library) user?

13

There are a number of emerging efforts in the library and information science community to use folksonomies within library catalogs. But folksonomies, just like controlled vocabularies, are only as good as those that develop them. If folksonomies are to be helpful then structure must be combined with diversity and ease of access. A brief look at four separate library OPACs subject headings and tagging use may be helpful in illustrating how this balance might be achieved. To simplify and provide only a sample, the bibliographic record of Suzanne Collins popular book The Hunger Games were compared. Subjective opinions were made about the usefulness or aboutness of each system. In looking at the Washington County Cooperative Library Services catalog (wccls.org) we found a website that uses only traditional subject headings and no tagging. This library system had the largest amount (9) and most descriptive subject headings. The second OPAC observed was the Ann Arbor District Library (aadl.org) catalog. This library system uses user generated tagging. In order to create tags, people must already have a library account in the AADL system. There are many tags in this system and a good number give more information about the book than the five limited subject headings provided. But many tags, like omg and 17_read5 are not helpful in describing the book. The third OPAC observed was the District of Columbias Public Library (dclibrary.org) catalog. This catalog had the same subject headings as the WCCLS system but also used their own users to tag items. In contrast to the AADL system there were very few tags (4) and they were not very helpful. This was surprising as the D.C. area has a population of over six times that of Ann Arbor. But this may have to do with how the tagging system was designed. As

CONTROLLED VOCABULARIES AND FOLKSONOMIES

14

Joshua Porter (2006) writes, If we create features just to aggregate them, without providing users with tangible value first, then people simply wont use the features. The fourth and final OPAC observed was the Deschutes Public Library (deschuteslibrary.org) catalog. Deschutes had the fewest number of subject headings used (4). Deschutes also employed a different way of folksonomy use. Instead of relying on their own users to populate records with tags, Deschutes used a service called LibraryThing for Libraries. This service pulls tags from the website librarything.com and displays them in library OPACS. The tags do not replace MARC records but is simply added to the displayed information in a bibliographic record. With over 86 million user generated tags and about one and a half million users, LibraryThing has a large base to pull from. LibraryThing then uses an algorithm that makes the OPAC only show the top tags in LibraryThing. This way tags like omg are vetted out ( LibraryThing for Libraries, n.d.). In looking at these four library systems OPAC catalogs we see that there is a wide variety in tagging use and effectiveness. The Deschutes Public Librarys adoption of LibraryThing for libraries seems to be the best balance between structure and ease of use. But there is room for improvement in many ways. Perhaps the most useful tool would be a way for users to add tags from library OPACs into LibraryThing. Then the tagging would be a two way sharing of information. Will Folksonomists replace the Experts? With the explosion of information access, creation, and sharing through the information age, it becomes harder for expert catalogers to keep up. More and more publishing is being done through blogs and other online publications. Yet there is limited money and time catalogers

CONTROLLED VOCABULARIES AND FOLKSONOMIES

15

have. Many proponents of folksonomies see their potential to help catalogers carry the workload and even extend the scope of cataloging. While it would be impossible for professional catalogers to catalog every website, folksonomists could. For every person that creates, there is also someone that can tag. Most in the field of library science see folksonomies as a field that is helpful. But they feel, as Lawley (2005) writes, that not all expertise can be replicated through repeated and amplified non-expert input. When used as a supplement to controlled vocabularies, they are great. But most contend that folksonomies cannot replace controlled vocabulary. This seems to imply a dangerous notion that folksonomies should remain subservient, or at most partners, to the experts. Conclusion What relationship folksonomies have to controlled vocabularies in the future will have a profound impact on how information is accessed in the future. Despite its relative infancy as a concept, it already seems that folksonomies and tagging are here to stay. Wikipedia.org was founded in 2001. Wikipedia is criticized because of its lack of authority. Yet Wikipedia is a non-profit, with administrators and mechanisms in place to reduce bias and increase reliability. Since 2001 nearly four million articles in English have been created by volunteers (Wikipedia contributors, 2012). By comparison, the latest (and last print) edition of the Encyclopaedia Britannica has only 65,000 article entries. While Britannica has about 4,000 contributors ( Encyclopedia Britannica vs. Wikipedia [INFOGRAPHIC], 2012), Wikipedia has about 300,000 active users contributing every month and nearly 17 million english language user accounts. Wikipedia is perhaps the most successful online model of a mix of many users, a

CONTROLLED VOCABULARIES AND FOLKSONOMIES smaller group of taggers, and an tiny amount of curators overseeing the project.

16

In the future, it will continue to be important that expert guidance and opinion are used in information seeking, creation, and access. While it is important that folksonomies themselves are improved; perhaps the biggest strength they have is in how they can complement conventional cataloging. If library goals continue to be the fostering of intellectual growth, then it would behoove the library community to embrace new avenues of information sharing, like folksonomies, and find ways to better integrate them. Librarians can react or they can adapt. Librarians should remain contributors in this collaborative project we call organization of information.

Appendix

CONTROLLED VOCABULARIES AND FOLKSONOMIES Fig 1: aadl.org

17

Fig 2: deschuteslibrary.org

Fig 3: wccls.org

Fig 4: dclibrary.org

References

CONTROLLED VOCABULARIES AND FOLKSONOMIES

18

AACR, J. S. C. for R. of. (2005). Anglo-American Cataloguing Rules. American Library Association. Deschutes Public Library /All Locations. (n.d.). Retrieved April 28, 2012, from http://catalog.dpls.lib.or.us/search~S16?/X(the%20hunger%20games)&SORT=D/X(the %20hunger %20games)&SORT=D&SUBKEY=(the+hunger+games)/1%2C22%2C22%2CB/framese t&FF=X(the%20hunger%20games)&SORT=D&7%2C7%2C Encyclopedia Britannica vs. Wikipedia [INFOGRAPHIC]. (2012, March 16).Mashable. Retrieved from http://mashable.com/2012/03/16/encyclopedia-britannica-wikipediainfographic/ Golder, S. A., & Huberman, B. A. (2006). Usage patterns of collaborative tagging systems. Journal of information science, 32(2), 198208. Guy, M., & Tonkin, E. (2006). Folksonomies - Tidying up Tags? D-Lib Magazine, 12(1). doi:10.1045/january2006-guy Holdings: The hunger games /. (n.d.). Retrieved May 5, 2012, from https://catalog.dclibrary.org/vufind/Record/ocn181516677 Lawley, L. (2005, January 20). social consequences of social tagging. Many-to-Many: Retrieved April 27, 2012, from http://many.corante.com/archives/2005/01/20/social_consequences_of_social_tagging.ph p LibraryThing for Libraries. (n.d.). Retrieved May 4, 2012, from http://www.librarything.com/forlibraries

CONTROLLED VOCABULARIES AND FOLKSONOMIES

19

Mathes, A. (2004, December). Folksonomies - Cooperative Classification and Communication Through Shared Metadata. Retrieved April 27, 2012, from http://www.adammathes.com/academic/computer-mediatedcommunication/folksonomies.html Merholz, P. (2004, October 19). Metadata for the Masses - Adaptive Path. Retrieved April 27, 2012, from http://www.adaptivepath.com/ideas/e000361 Peterson, E. (2006). Beneath the Metadata. D-Lib Magazine, 12(11). doi:10.1045/november2006-peterson Porter, J. (2005, March 9). Controlled Vocabularies Cut Off the Long Tail - Bokardo. Retrieved April 27, 2012, from http://bokardo.com/archives/controlled_vocabularies_long_tail/ Porter, J. (2006, May 2). The Del.icio.us Lesson - Bokardo. Retrieved April 27, 2012, from http://bokardo.com/archives/the-delicious-lesson/ Shirky, C. (2005). Ontology is Overrated: Categories, Links, and Tags. Clay Shirkys Writings About the Internet. Retrieved from http://shirky.com/writings/ontology_overrated.html Sinha, R. (2006, January 6). A social analysis of tagging. Rashmis blog. Retrieved from http://rashmisinha.com/2006/01/18/a-social-analysis-of-tagging/ Speller, E. (2007). Collaborative tagging, folksonomies, distributed classification or ethnoclassification: a literature review. Library Student Journal, 2(0). Retrieved from http://www.librarystudentjournal.org/index.php/lsj/article/view/45/58 Spiteri, L. F. (2007). The Structure and Form of Folksonomy Tags: The Road to the Public Library Catalog. Information Technology & Libraries, 26(3), 1325. The hunger games | Ann Arbor District Library. (n.d.). Retrieved April 28, 2012, from

CONTROLLED VOCABULARIES AND FOLKSONOMIES http://www.aadl.org/catalog/record/1319099

20

Trant, J. (2009). Studying social tagging and folksonomy: A review and framework. Vander Wal, T. (2007, February 2). Folksonomy:: vanderwal.net. Retrieved April 27, 2012, from http://vanderwal.net/folksonomy.html Washington County Cooperative Library Services. (n.d.). Retrieved April 28, 2012, from http://catalog.wccls.org/polaris/search/searchresults.aspx? ctx=1.1033.0.0.1&type=Keyword&term=the%20hunger %20games&by=KW&sort=RELEVANCE&limit=TOM=*&query=&page=0#__pos8 Wikipedia contributors. (2012a, May 3). English Wikipedia. Wikipedia, the free encyclopedia. Wikimedia Foundation, Inc. Retrieved from http://en.wikipedia.org/w/index.php? title=English_Wikipedia&oldid=490276746 Wikipedia contributors. (2012b, May 4). Tag (metadata). Wikipedia, the free encyclopedia. Wikimedia Foundation, Inc. Retrieved from http://en.wikipedia.org/w/index.php? title=Tag_(metadata)&oldid=482365846

You might also like