Ali Shiri and Dale Storie, University of Alberta
[Please note: These are merely my notes on the presentation, taken live while the presentation was in progress and edited for sense afterwards. They are not a verbatim transcription of the presentation, and any errors are mine. Please contact the researchers directly for more information on their work. — JRD]
Notes on the presentation:
Social tagging has been referred to by a variety of terms. It’s an area that’s emerging. The process is basically that the users interact with a variety of web resources, whether it’s textual information, multimedia, journals, books, etc., and assigns tags. They could be called tags, terms, keywords, etc. The result of that sort of tagging can be represented in a variety of formats. Tag clouds are one type of representation of tag space and can actually let a user refine or modify their query.
Our study was exploratory, meant to explore, identify and categories interface features and functionalities that social tagging sites offer to allow users to create, contribute, explore and interact with content, particularly tags. Are there any features specifically designed to influence users’ tagging behaviour? What types of tag representation features can we find on the page of a site?
Shiri, ECDL, 2007
We selected 10 tagging sites that were popular, widely used, and include both social bookmarking and social media. 4 selected sites were social media, and 6 were social bookmarking.
We were particularly interested in discovering whether there might be implications for controlled vocabularies.
- Social media: youtube, myspacetv, bubbleshare, flickr
- Social bookmarking: delicious, furl, connotea, citeulike, technorati, and backflip
We’re aware that technorati is a ‘blog search engine’ but they use social tagging in a variety of interesting ways.
Analysis Framework
We decided to take an analysis approach rather than a theoretical one. We looked at user tagging features, exploration features, interface layout, and the content type and tagging features provided. Some sites allow tag grouping, browsing popular or recent tags, using system suggested tags, etc. We intentionally used the term ‘exploration’ to cover both browsing and other kinds of interactive searches. The term ‘exploratory search’ has become popular in the literature for information discovery. And page views for browsing might be incorporated right into the home page. And is there any specific pattern, or are there specific features for social media sites as opposed to social bookmarking sites?
To reinforce the idea, tagging is related to informational organization, while tag browsing relates to information interaction.
In terms of tagging features we looked at tagging type (broad or narrow), the number of tags allowed, etc. Social media sites tend to use narrow tagging. A person who shares a video file can only assign tags to that file, for instance.
We found that five sites used preset categories of tags, which is a sort of a shift back to hierarchical structuring of data. Also, not all sites offer multi-word tags, which can be a limitation for those services that don’t offer that sort of facility. Six sites offer recommended tags; we were looking at this from an information organization perspective.
Social media sites tend to use pre-defined categories, while social bookmarking sites tend to use free tagging. Mainly bookmarking sites use multiword tagging; both media and bookmarking use tag recommendations. Connotea and backflip use tag notes, which is interesting because this is like a scope note in a formal data organization system.
Tag bundling and grouping is another shift toward a hierarchical structure; you can group in terms of facets.
Geotagging refers to location based tags, cities, countries, etc. Flickr, delicious, and connotea use this.
YouTube implements predefined categories; it ranges from five to 20 depending on the site. Furl offers topics, which basically means predefined categories; you can see that the tagger can choose any or all of these categories depending on the subject or aboutness of the content or site they are sharing.
This is bubbleshare, a photo sharing service. They have predefined tags, and users are sort of forced to categorize or they are free to tag but there is some behind-the-scenes work to put themm into those categories.
This is a geotagging feature in flickr. Users can assign location-based features. There are metadata elements that have been incorporated into social tagging sites.
Connotea lets users add a tag note to explain tags in more detail.
Looking at the tag browsing features, users can browse by tag, popular, recent, related tags; automatic grouping of tags (flickr only), and delicious and flickr let you browse featured tags. all social bookmarking sites let users browse users’ tags but only Flickr lets users do this on the media tagging sites. Tag clouds which are becoming more popular on PACs, etc. were present on 5 sites.
I’ll talk about automatic grouping and related tags later on.
In terms of information interaction there are different features: tag lists, tag clouds, and individual tags. We’re just focusing on one aspect of this. Examples include featured tags, automatic clustering, browsing users’ tags, browsing posting history (delicious and connotea), allowing people to browse the collaborative intelligence of the users.
CiteULike offers a taglist, a collection of tags for browsing. On delicious users can choose from either popular or recent tags. Connotea uses recently used and related tags (based on automatic clustering). Technorati uses tags, search terms,and predefined categories. Youtube is based on individual tags and they’re based on a description of the actual resource.
Delicious offers an interesting tag cloud; it can be organized alphabetically or by size; red tags are the ones you share with other people. Flickr has a combination of tag clouds by time/popularity. If you put in a term like “turkey” in flickr you can see they’re clustered for disambiguation purposes.
General observations: Delicious and citeulike have the most tags on the homepage; some services make it easier to access tags than other; some tags use tags more for searching, and some more for browsing.
Services with information organization purpose tend to offer more tagging features.
Social tagging focuses on personal, collaborative, and social organization factors. Users tap into collective intelligence using many types of tag interactions, and using both conceptual and visual capacities.
Future research will focus on how controlled vocabularies can be utilized in tagging settings. Some OPACs have LCSH headings in a tag cloud format.
Question from the audience: I was thinking, why haven’t we done this all these years, if users would use index “clouds”? But I think these sites are more entertainment based than information based. Do you really think people will use this to find information?
Answer: I’m doing a literature review now. Some people use browsing rather than searching, and vice versa. I’ve seen mixed results. Some of it is related to web development; tag clouds are a new feature and they can be used nicely. I’ve been working on these projects for the last five years that if you provide categories or faceted browsing, it can be useful. But you’re right, it should be subject to actual user study to see if they would like to use it in a non-entertainment context. But there are lessons that can be learned and applied.
Moderator comment: People say they don’t use tags, but they often do, in studies that I’ve done.
Question from the audience: Who decides what is the ‘important’ way to represent tags? How can you use this in a controlled vocabulary situation when the users are the ones more relied on for determining these features?
Answer: I’m not saying we should be using all these features. But a large percentage of North American libraries are using LCSH. How can we explore using some of these technologies to enhance the use of these systems? There are lessons we can learn from social tagging without changing our subject vocabularies. There are some OPACs that are adding social tagging. It’s not transforming the structure, content or semantics of controlled vocabularies.
Question from the audience: I think 20 or 30 libraries now have integrated LibraryThing seamlessly into their OPACs. People don’t know that it’s LibraryThing.
Answer: You’re right, social tagging is being used in a variety of contexts. University library portals are incorporating social tagging into their features. it’s becoming more popular, but since LT is sort of a classification structure as well … but we haven’t looked at that for this particular study.
Question from the audience: Are we comparing apples and oranges here in using Knowledge Organization terminology to look at something that isn’t search and retrieval? Maybe it’s about interaction, sharing, curiosity, and exploration, that has nothing to do with search and retrieval in their traditional forms. If we’re doing user studies, maybe we should look at what people are doing in their real lives, not just saying ‘how do you find stuff’?
Answer: There are projects looking at re-using controlled vocabularies in a way that’s more suited to the web environment, like using XML. How can we use these strong knowledge structures with a rich semantic structure, like LCSH, with new approaches that might make them more usable and useful? It might be a way to make controlled vocabularies more visible to the users.
