Heather D. Pfeiffer, New Mexico State University
Emma Tonkin, UKOLN, University of Bath
Mark R. Lindner, University of Illinois at Urbana-Champaign
David R. Millen, IBM T.J. Watson Research
Margaret E.I. Kipp, College of Information and Computer Science, Long Island University
Heather D. Pfeiffer – Tagging as Metadata
Heather talked about knowledge in language and how we understand syntax (symbols), semantics (meaning), and pragmatics (context) in that language. She mentioned the conceptualization of an ontology and how relationships can be drawn between concepts. Which lead into concepts as tags. There was an interesting comparison example between documents in the 1600s and in the 2000s.
1600s Document’s terms
||2000s Documents terms|
In the 2000s the word “upon” is not used, King has been updated to President, and Crown appears in both. In the 1600s Documents Crown refers to “Swedish money” and to government (in relationship to King), in the 2000s Documents Crown continues to refer to money but there is no longer a relationship to government because of the shift to President instead of King.
Emma Tonkin – Ten Minutes of Language Development
Emma started with a little exercise by naming places from “The Atlas of the True Names of Places”, something she picked up in her travels. Some of them were fun and interesting like “Empty Place”, “Little Venice”, “Land of People Who Are Their Own Masters”, and “East of the River of Shellfish”. She mentioned a three-step ontology development plan which was to identify concepts, label concepts, and identify relations. The assumption is that this is an easy process, but realistically identifying and discussing labels is hard. She asked us to consider the following; 350 N High St is important to us now, because we are at this conference, but then also consider place vs space and position vs. location, labels like ‘home’ and ‘office’ mean things to us personally but meaningless to others.
I actually find this discussion interesting, but less so in the context of tagging and labeling and more so in social contexts of geo-location. But I can see how these labeling concepts can apply in the tagging discussion.
She had some good examples in her presentation which are difficult to parse out here, but she presented us with a game board and how the labeling of things depend on a variety of variables, including location and context in language. The example she gave was pharmacy (UK) and drugstore (US) and how this could be what she called a Deadlock Situation if there’s an even representation of people using that language or labeling. Whereas if you’re in America this same label could turn into a Majority Voting situation if you say had more representation for drugstore than you do pharmacy. What is the probability that we come to a consensus? She mentioned that we actually don’t want perfect accuracy. What happens if we moved the drugstore location? The labeling problem is that “if nothing ever changes, nothing interesting ever happens.” Which means this is a “living system.”
David R. Millen – Patterns of Collaborative Tagging in a Large Organization
David started with some background about the project and how they went about studying tagging in the enterprise. He also gave some background on Dogear, IBM‘s internal social bookmarking tool. Their goal was to examine enterprise as a group, and IBM is a good company for this because it has the tools already in place plus they are very large, global, and multilingual. The research group looked at three different core groups when they studied tag use; Development, Research, and Corporate. They observed broadly, and within these groups, that there was more similarity than differences. The number of unique posters is about 50% and relatively steady across the 3 groups. They are looking at this number as a metric or a “return on contribution”. They want to know if that ratio stays the same across all 3 groups.
The research group studied tags per bookmark and also observed tagging on the internet versus the intranet. What they found was that there were more tags on the intranet, which wasn’t so much what they expected. They think that this may be because of the homogeneity of the intranet, or similar kinds of resources.
They studied frequency of tags and classified these into categories of topic, content, and owner. They found a lot of tag consistencies and within the 3 groups of study they found consistency to what they would expect the group to be interested in.
Then they also studied private bookmarks, what they found was that the amount of private bookmarks that are tagged is stable and low. What they were able to determine from this, and he said this is “good news”, was that people are sharing.
He summed up by stating that tags are good markers of organizational interest in a large group and he also mentioned that tagging in IBM is not isolated to Dogear, there is tagging going on in other social spaces withing the organization.
I found one of the more interesting areas of the presentation to be the “roles” of taggers in the organization. He said they’ve determined 5 roles, but went through 3 in this session.
- Evangelist: These taggers are trying to cultivate an audience on a term, they want people to find them so the tags are also self-serving.
- Publisher: These taggers want to draw attention to the content and bring people to resources.
- Small Team Leader: These are conventional tags used by small groups so they can easily find resources. They have determined that these taggers are less active.
He also mentioned Wordle yea! Here’s mine:
Mark R. Linder – Integrating Tagging: Tagging as Integration
I admit I didn’t take much notes here, more on tagging in language, context, and use. There’s a couple of resources listed in the proceedings on information seeking, semantics and knowledge organization.
Margaret E. I. Kipp – Social Tagging Process
Margaret spoke on the process of tagging and trying to figure out what something is about. The process of tagging seems simple on the surface, but figuring out what something is about can be a little more complicated. Tagging looks like a classification index on the surface so it’s natural to want to compare it to that. When you look at a consensus graph on the Clary Shirky’s article, Ontology is Overrated, you see that the most popular term is ontology. The first 6 terms are subject related, followed by the name of the author, then followed by a tool. Maragaret broke down different uses of classification:
- User Classification: The author and user are placing different contexts on the article.
- Information Management: The user may be tagging something to find again later.
- Communication: This is where you see more non-subject terms like ‘todo’, ‘toread’, ‘funny’, and ‘cool’. Also tagging and for small groups appears, which may indicate a community.
- Discussion of Aboutness: There is a fair amount of agreement when put together with the whole group, but as you look at individual taggers you do see some disagreement.
- Expression of Interest: Tagging in itself is an expression of interest, if you weren’t interested you wouldn’t tag it. But tagging can also indicate a degree of interest, to read, or to buy later, for example.
- Review or Criticism: Terms like boring, fun, or funny indicate some emotional reaction to an item. This becomes and ultra condensed review and mixed with other terms might be useful to the somebody else.
- Projects and Groups: Tags from CiteULike that indicate course numbers or groups for communicating resources are an example.
- Time Sensitive Classification: The term ‘toread’ represents this, the meaning can change over time. Once you read and item the tag becomes irrelevant, or maybe you never read, or you read and don’t remove it later.