The Synthetic Librarian

Social Bookmarking is popular and remarkable because it not only allows users to record the addresses of their favourite websites, but because it allows them to share descriptions and classifications of those websites. The sharing of information about “what’s good” on the web, elevates sites like delicious.com, Cite-U-Like, Digg beyond being mere “bookmarks” into being information discovery services. Libraries should be bold adopters and advocates for this technology, and become its experts users; for long before these services existed, libraries were the social bookmarking service.

It is trivial to claim that libraries are in the business of describing, classifying, and organizing information (primarily books and printed periodicals). It is more controversial to claim that the way libraries do this is similar to a social bookmarking service. At their best, libraries are much more than organized warehouses for books, they are information ecosystems whose environment is constantly under modification by its inhabitants. The discover of resources in a library is enabled and enhanced by the presence of its users and enhanced by the day-to-day activities that occur within it: that is, the books gets selected because of the participation of people, not just because of their place on the shelf.

Beth Jefferson of Bibliocommons is fond of giving a classic example at library conference presentations. Users have long used the return cart at the library as a method of discovering “what’s good”. If a book is sitting on the return cart, it stands to reason that someone else took it out and that itself stands as a recommendations. I have personally heard stories of libraries that used public sign-up sheets for putting holds on popular items being used this way by patrons: the books worth reading on the ones that other people are asking for! This is what social bookmarks do too: show you what other people have read. This is just one example of the kinds of unseen “discoveries” that take place in a library. The reference desk is a more obvious source of social discovery, as is the library-support reading group, or chance meetings that occur between patrons on any given day. Of course, this is not how information discovery is emphasized in a library. Instead we talk about information retrieval by searching, by authority control, and strict classification. Yet, realistically, users and their social interactions have an enormous impact on what is discovered.

Social Bookmarking is not entirely like libraries of course: it relies for more heavily on the participation of users to enable resource discovery than libraries ever have. If you imagine the situations that might take place in physical libraries that I just briefly outlined, you must imagine that “social sharing” takes place slowly: any individual experiences these sharing experiences once or twice per visit to the library. The wisdom of individual librarians is available to everyone, but the wisdom of everyone else is only indirectly available.

Social Bookmarking sites change this by making the recommendations and perspectives of others available directly and in aggregate. In the return-cart example, knowing that a particular book had been borrowed by one other person might be taken as an implicit recommendation of that book over one still on the shelf (and thus probably less popular) but if you could know which books were read by 100 other people, you’d have a stronger recommendation.

The Elements of Social Bookmarking

Social Bookmarking, from the librarians perspective is an information retrieval system with several key characteristics:

• URLs (webpage addresses) are the unique resource identifiers
• Title of a work is probably recorded but author and date are not significant
• Classification does not use a controlled vocabulary
• The identity of who bookmarked the URL is recorded

These characteristics contrast greatly to what is considered essential to information retrieval in libraries and yet social bookmarking is still popular and effective. Fundamentally, the concept of relevance is at the heart of both libraries and social bookmarking. “Information retrieval (IR) has the objective of obtaining relevant documents from document collections given queries provided by users” (p. 107). Library classification strategies work because the number of classified items are relatively small. A small group of experts can determine which items are relevant to a broad set of general queries. Social Bookmarking applies to a completely unconstrained and vastly diverse collection. Classification of this diverse collection is solved with diversity itself. No assumption is made about who is or what an expert might be on items to be classified. Instead, the assumption is that some individual will have already discovered items that are relevant to them. If those individuals share their classification of these items, other people with the same perspective or needs will then find those items as well: everyone is an expert on what they have already found to be relevant.

How should Libraries View Social Bookmarking: What does it do for me?

Enhanced Search

First and foremost, staff who work in libraries must be prepared to use social bookmarking sites as a new search enhancement tool. This is not just because of the popularity and hype, but because early research indicates that social bookmarking is and will be an enhancement to search.

For example, Yanbe, et al. (2007) performed an analytical study comparing to search algorithms, one using a traditional ranking mechanism, and one using a ranking mechanism enhanced by social bookmarking data. They found that relevance of search results to user queries was enhanced by including social bookmark data. Search engines that incorporate this information are not yet available, but the expert searcher (i.e. You, the library worker) can already do this. For example, using the results of a google search to inform choices for a search of delicious.

Example, step-by-step: Follow these steps to effectively combine a Google and a Delicious search.

1. Perform a query in Google as usual.
2. Copy the URL of a relevant result
3. Bookmark this in delicious
4. Now, pull up your delicious bookmark for that URL, and find out who else bookmarked it
5. Now, find out what some of those people tagged it as. Make note of the tags that you think are relevant
6. Now, browse the other links that those people tagged with the relevant tags
7. You have now enhanced your search abilities, and possible discovered things that Google would never have shown you.

This is a bit tedious, but its fun. I’m confident that in no time at all, this will be incorporated directly in to search engines so how.

Leadership in Folksonomy

A key characteristic of social bookmarking is the lack of external authority control on classification terms. There is not standard vocabulary. This gives people the freedom to pick terms that are as specific or general as their needs. However, it also means that misspellings, acronyms, and other idiosyncrasies can impact the “shareability” of items. Librarians are in a unique position to improve this situation. First, librarians can lead by classifying web resources using better indexing terms than others. Second, librarians can establish and advocate for improved folksonomy practices.

Librarians are represented and participate in an extraordinary number of committees and professional associations and these are active in settings standards and advocating for improved practices among librarians. In this case standards are not desirable or required, and the practices that need improving are global in scope. However, I believe that librarians have the opportunity to take a leadership role in establishing improved folksonomy practice.

Advocating Social Bookmarking

Older (2008) interviewed a librarian from Oakville Public Library, one of the first libraries to deploy Bibliocommons, a social discovery system for Integrated Library Systems (ILS). In that interview social discovery is called “revolutionary” by a librarian who has used it. Bibliocommons is a walled garden approach to social enhancement: user contributions are shared only with patrons of other libraries using the Bibliocommons system. The potential for open social bookmarking is tremendous but relies on similarily tremendous user participation. Libraries could play a key role in advocating for the use of social bookmarking in general.

It is easy to see how, when asked for advice from a patron, on ways to manage the overwhelming amount of information on the Internet that they might advocate a social bookmarking solution. The need however may be greater. Benbuann-Fich & Koufaris (May 2008) examined the motivations of users in social bookmarking and found that “users contribute tagged resources for other users only if they believe they will be useful for those users. Moreover, higher quality contributions for others do not diminish the quantity of such contributions. We also find that there is a spill-over effect from quality of contributions for self to quality of contributions for others” (p. 150). Thus there is a snowball effect to be had. Users will contribute more if they feel others value it, and quality increases by some encourage quality increases by others.

Tagging information resources with keywords has the potential to change how we store and find information. It may become less important to know and remember where information was found and more important to know how to retrieve it using a framework created by and shared with peers and colleagues. (EDUCAUSE, 2005).

Librarians have a natural role in helping individuals to discover information tools that will help fill their needs, but with social bookmarking, such advocacy could result in entirely new synergies.

Where is Social Bookmarking Done

Social bookmarking is much more widespread and more diverse that one might realize. The common examples are “pure” social bookmarking systems like Delicious, Digg, and Reddit. With this systems users bookmark, describe, and classify individual URLs. However, there is remarkable new areas in which social bookmarking is appearing.

Elements of Social Bookmarking are Widespread

One would be incorrect to call Flickr or Youtube social bookmarking systems, however, these system do contain important aspects of social bookmarking features. For example, both Flickr and Youtube allow users to mark their favorite photos and videos (respectively), and allow others to view a user’s list of favorites. Importantly, this allows for the same kind of resource discovery enabled by social bookmarking services, but only within the boundaries of a single website.

This begs the question, “why do social media sites implement their own bookmarking? Why not implement bookmarking as a mashup with a social bookmarking service?” Certainly this would provide more value to the end users. One reason, in theory, is that custom bookmarking within the site allows for enhancements specific to the type of media being bookmarked.

For libraries, I believe that the clear choice is to implement social bookmarking as a mash-up with one or more 3rd party systems like delicious.com, cite-u-like, etc. As an example, the University of Alberta’s catalogue offers minimal integration with several social bookmarking systems. On the left-hand side of the details page for a specific book there are links to delicious.com, reddit, digg, and google bookmarks. Users will find however, that if they attempt to use standard bookmarklets or toolbars for social bookmarking sites, that the bookmarks do not work at later dates. This is because the University of Alberta’s catalogue puts random ‘session information’ in the URL of every page. This information is different every time you visit the site, making it impossible to bookmark any page. It takes special technical tricks to enable any integration with social bookmarking at all.

Social Bookmarking for Scholars

Of particular interest to library staff are social bookmarking services that cater to the needs and items of interest to scholars. Cite-U-Like, Connotea, and GetCited are all social bookmarking services for bookmarking scholarly articles. [Full Disclosure: I believe GetCited was developed in part by Emergence by Design, a company that I once was a partner in. It was developed after I divested my interest in the company.]

These social bookmarking services differ from the classic ones in that users are not bookmarking URLs but scholarly works. The systems know, at a technical level, how to identify the work from the URL being bookmarked, and will extract substantially more metadata than, say, Delicious would from a webpage. This can lead to richer collaboration, sharing, and discovery, between scholarly users of these systems as the data is more suited to their needs.

In the next section, I will describe my own experience using Cite-U-Like.

Exploring Cite-U-Like

I am already a heavy user of the delicious bookarking system. So for me, using Cite-U-Like definitely provokes comparison.

First, Cite-U-Like offers me many features that I value. I do often bookmark scholarly articles but I tend not to use delicious for that purpose because the way I search for such articles depends on specialized metadata unique to scholarly articles (e.g. title, ISSN, volume, issue, doi, etc.) I use Zotero to bookmark and find articles, and before that I used Refworks.

To begin my exploration, I exported my old RefWorks citations in BibTex format, and then imported those into Cite-U-like. This immediately established a collection of bookmarks for me to explore. My bookmarks in Cite-U-Like can be see publicly at http://www.citeulike.org/user/clonedmilkmen.

I notice that the import feature is somewhat imperfect in that it created tags from sentences describing the post. Thus, I ended up with many useful and not so useful tag (e.g. “and” is a tag). However, overall, the import process was useful.

To add new bookmarks, I prefer a method similar to what I use with Delicious. Currently, I use the delicious toolbar buttons for Firefox. To bookmark a page, I click “control+D” and a window opens allowing me to enter in information. No toolbar exists for Cite-U-Like but they have a “bookmarklet” that I can add to my bookmarks. When I activate the bookmarklet a pop-up window opens asking me to describe the item.

Unfortunately, the bookmarklet did not work for me. To test, I would search a database, such as Academic Search Complete and select an article to view. Then I would activate the bookmarklet. This should have parsed information about the article to be bookmarked, saving me the work of having to type it in manually. However, each time I tried this I was presented with the message “Unfortunately, we could not process the URL you submitted….” and a form that I had to fill in myself.

The form itself presented much richer input options than delicious (which offers title, description, and tags).

It is worth noting that one can bookmark general webpages with Cite-U-Like. General webpages are supposed to generate the “Unfortunately, we could not process the URL you submitted….” error by design as there is no scholarly metadata to parse.

I found using Cite-U-Like for discovery of resources to be a mixed experience. On the one hand, it seemed that the tags other used, and the things they bookmarked were highly specific to their research. So, general tags like “security” or “information security” yielded things that were completely irrelevant to me. The same type of search in Delicious has the opposite results: lots of interesting things. On the other hand, I was able to quickly find other people with similar interests and discover articles by looking over their entire bookmarks. I found that most users had a small number of bookmarks in total but that may be because I have not had enough time to explore further.

It is clear that Cite-U-Like is in an interesting niche and after my experience, I believe that my advice above that librarians have a role to advocate for better practices in folksonomy are doubly important. In Cite-U-Like, with few users, with more specific interests, general tags are not useful, and specific tags must be applied with care. Expert advice would benefit the community.

I think I will be inclined to use this tool much more. In particular, I think I would like to see tight integration with Zotero. I would love it if I could bookmark and describe items in Zotero but share and discover them through Cite-U-Like (e.g. automatic posting of my citations from Zotero to Cite-U-Like). Since that does not exist yet, I will experiment in the future with exporting my Zotero database and importing it periodically into Cite-U-Like.

Social Bookmarking and Privacy: Isn’t what I read supposed to be private?

I would like to conclude by commenting on the implications of social bookmarking for privacy. Libraries have a very special relationship with privacy. In order to make all users feel comfortable and safe reading and seeking information, libraries must give patrons their privacy. There are books patrons would not ask for, and questions they would not ask, if they believed their interests would be made public. So, we then must ask, isn’t asking people to share what they read in contradiction to a library’s commitment to patron privacy?

To answer this one must consider that privacy is not the same as confidentiality or secrecy. Confidentiality and secrecy are often how we guarantee privacy. Privacy is an individual’s right to control information about them (Acquisti, et al., 2007, Chapter 1). We are no violating patron privacy by encouraging them to share what they read, in fact, we are closer to violating when we deploy library systems that given them no choice to keep what they read secret. Giving individuals a choice about what they share, how they share it, and with whom they share it, is a privacy practice that is empowering as well as respectful.

Librarians must become experts in social bookmarking then and experts advocates as well. They must be prepared to use social bookmarking to find materials, elevate the usefulness of classification being used, to educate users to make the best use of it, to enable the private use of it when it is combined with library systems.

References

Acquisti, A., Gritzalis, S., Di Vimercati S. (2007). Digital Privacy. CRC Press

Benbuann-Fich, R. & Koufaris, M. (May 2008). Motivations and Contribution Behaviour in Social Bookmarking Systems: An Empirical Investigation. Electronic Markets, 18(2), pp. 150-160.

EDUCAUSE (May 2005) 7 things you should know about social bookmarking. EDCAUSE Learning Initiatiive.

Older, N. (2008). Bibliocommons Emerges: “Revolutionary” Social Discovery System for Libraries. Library Journal, 07/18/2008

Jefferson, B. (2007). Forget the Lipstick. This Pig Just Needs Social Skills. Code4Lib Conference, 2007 (video).

Puspitasari, F., Lim, E., Chang C., Theng Y., Goh, D., Chatterjea, K., Zhang, J., Sun, A., & Li, Y. (2007). Social Bookmarking in Digital Library Systems: Framework and Case Study. Joint Conference on Digital Libraries, June 18-23, 2007, Vancouver, British Columbia, Canada.

Yanbe, Y., Jatowt, A., Nakamura, S., & Tanaka, K. (2007). Can Social Bookmarking Enhance Search in the Web? Joint Conference on Digital Libraries, June 18-23, 2007, Vancouver, British Columbia, Canada.