Fading Waypoints

first post from performancing, which i hope will make my blogging tasks more managable.

We will be exploring collaborative tagging in our workshops, and have been looking for interesting examples to guide the design of them. Anton found STEVE earlier and going through some of their references i was happy to see The ESP Game mentioned.  Such a simple HHCI , yet so much data was gleaned.

We began designing our own workshops to be a much lower fidelity, (think post-its) but i think some game element, even a superficial one might drive user participation. In applying rules to the act of tagging, a game provides lots of opportunities to design relationships between users, the information, and the interface.  Imagine if you will, (massive?)multi-player Tic TAG Toe. (try this classic version for now, http://www.tictactoe.com/)

Tic Tac Toe is a Zero Sum Game,  and as i surf through the wikipedia links this could be a can o worms.

An interesting pattern keeps appearing in my googling for tagging workshops that may lead to some interesting discoveries - Animal Tagging.
Animal tagging shares at least one goal with information tagging - findability.

I don’t want to get too distracted by this right now, but I thought I would throw it out there.

I am currently doing the Google rounds on existing (mob) tagging workshops in order to prepare for our workshop/paper and I came across steve - Social Terminology Enhancement through Vernacular Engagement.[link]

Steve is a project being developed by a group of museums that want the public to engage with their collections by allowing them to search for an item though a social filter. The way museum curator catalogues a collection is rarely the same as member of the public would search for it.

Curators rarely answer the question “What is this a picture of?”[link]

For example:

A curator may catalogue this image as Italy,futurism, painting, 1912, Balla. Giacomo
Whereas a person searching for it my use terms like dog, daschund, motion, blur

Steve is an attempt to democratize a collections catalogue, not by replacing the curators perspective, but by adding an additional layer of information provided by the public.

Bonus - Steve appears to be under a creative commons/share and share alike licence.

This local blogger identified some of the false hopes of technorati tags back in september. He questions…

Why do I need words to describe the words I’ve already written?
Why must I artificially categorize my text content?

These are very pertinent questions to our research and as we begin designing workshops for users to explore the concept of tagging, we should keep them in mind. We believe there will eventually be a more dynamic relationship between user generated tag spaces, and the supporting software architecture that works behind the scenes.

Darren continues later with the call to arms…

Let’s get busy teaching our computers how to differentiate this apple and this Apple.

It seems teaching this information to a computer, or designing an ontology that can do so is quite difficult. If users can conform to some simple rules, perhaps a more descriptive use of tags can emerge.

We should also think about addressing some of the comments from our peer review, as well as some well known issues with tagging and folksonomy in general.
Some points that I think we need to follow up on:
Natural Language Part-of-Speech Tagging. Hidden Markov Model
Synonym & Homonym problems.
Hidden Long Tail - Chris Anderson
Tags can be temporal - and subject to change: toRead, toSee, toBuy, toVistit; read, seen,  bought, been

A quick brain dump that I intend to revist, edit and breakout from as I see fit.

One of the first things I think we need to do is generate some data on tagging. In my opinion, the best way to do this is to get a bunch of people to tag a bunch of stuff a mobtagging party or something. Invite your friends - it will be cool. You do want to be cool, don’t you.

The purpose of the flashburningmobtagsonomic workshop is to get a sense of how people tag things, how the act of tagging effect the taggers relation to the data, and the data’s relation to other data - how the data organizes itself).

I would like to implement a couple of iterations on this process.

1. Free Form Tagging.
2. Weighted tagging - manual weighting - like setting a rating in itunes.
3. Weighted tagging - position sets the weight - first tag is more important than the second etc…. if two or more tags share the same weight, then place them in square brackets [ ]

Questions:

• What exactly should we have people tag? Books? Movies? Music? Recipes?
  My feeling is we should stick to things that are easy enough to categorize on their own and something people may feel passionate about.
• How many things do we need in our data pool?
• How many people should we get for the workshop?

Some times people feel compelled to create their own landmarks in places they have been. Inuksuk and grafitti are two of my favorite examples. These serve as physical reference points for the individual who creates them and others that share the space, declaring i was here, or this is the way, among many other messages. They communicate to all others who pass that location.

What interests me, is that the same user behaviour can be observed on the web, for example by users saving, and tagging links. Recently,this abilty for users to leave personal landmarks, or tags for all to see has become popular. Often referred to as folksonomy, a socially constructed network of metadata is evolving and becoming a dominant force in web navigation, search, and communication. Whats going on here? What compels users to input ‘keywords’ of associated meaning alongside specific content on the web? What is the ROI?

Although users have their own intentions and context for tagging content, it is becoming clear, tags have a life of their own and as a collected set, new meanings can be observed. Anthony and I wish to explore this idea a little further.

IA Summit 2006 Reviews for Submission #183

Title:The Life of Tags
Authors: Anthony Charles & Jason Toal

REVIEWER #1

Reviewer’s Scores

Relevance to Information Architecture: 6
Contribution to Information Architecture: 7
Originality and creativity of ideas: 9
Quality and rigor of research or design approach (for research posters): 5
Clarity and organizatrion of presentation and writing: 3
Potential of Topic for Engaging Audience: 7
Potential to Aadvance the state of the art: 6
Significance of findings: 6
Overall Rating: 5
Recommend: Prob reject (boo)

Detailed Comments

Combining the analysis of tags and CA is an interesting idea. For the research track we are more inclined to include actual studies that address your ideas. It is possible to build a paper around the theory of your ideas (the rules to the process of tagging), is that a direction you will be pursuing? What data, empirical analysis or previous studies could you use to fortify your points?

(Note: some recent talk about tagging makes the argument that rules for tagging are against the philosophy of tagging. Tag assignment should be ad hoc and not based on any formal rules, as they would either be ignored or performed inconsistently (much like indexing) so much so that their supposed consistency isn’t good enough to be depenedable for a system relying on the tags.

============================================================================
REVIEWER #2
============================================================================

Reviewer’s Scores
———————————–

Relevance to Information Architecture: 10
Contribution to Information Architecture: 10
Originality and creativity of ideas: 10
Quality and rigor of research or design approach (for research posters): 11
Clarity and organizatrion of presentation and writing: 10
Potential of Topic for Engaging Audience: 10
Potential to Aadvance the state of the art: 8
Significance of findings: 8
Overall Rating: 10
Recommend: accept

—————————————————————————
Detailed Comments
—————————————————————————

As my ratings reveal, I liked this proposal a lot. Even though the analogy between user-tagging content and cellular automata systems might be a bit of a stretch, I like the novel approach and effort. It will be interesting to see whether the author(s) have a workable method for writing rules about natural language behavior! It’s an ambitious project and one that is likely to be thought-provoking even if the outcome falls somewhat short of the promise.

============================================================================
REVIEWER #3
============================================================================

Reviewer’s Scores
———————————–

Relevance to Information Architecture: 7
Contribution to Information Architecture: 6
Originality and creativity of ideas: 6
Quality and rigor of research or design approach (for research posters): 3
Clarity and organizatrion of presentation and writing: 5
Potential of Topic for Engaging Audience: 5
Potential to Aadvance the state of the art: 6
Significance of findings: 2
Overall Rating: 5
Recommend: marginal

—————————————————————————
Detailed Comments
—————————————————————————

Tagging is evolving and the evaluation of tagging is equally important to our field. The subject matter is relevant and has potential to contributing to the field, but there’s no mention of findings, which makes it difficult to judge how relevant it really is to the field. It could be, if presented correctly, but may not be if there are no findings. Either way, even the theory could be beneficial to the field.

Tags are something we’ll all need to deal with. So, I would welcome something on this subject, it just needs some more work. Hope to see this polished up by the summit.

============================================================================
REVIEWER #4
============================================================================

Reviewer’s Scores
———————————–

Relevance to Information Architecture: 7
Contribution to Information Architecture: 8
Originality and creativity of ideas: 7
Quality and rigor of research or design approach (for research posters): 5
Clarity and organizatrion of presentation and writing: 6
Potential of Topic for Engaging Audience: 7
Potential to Aadvance the state of the art: 7
Significance of findings: 6
Overall Rating: 7
Recommend: prob accept

—————————————————————————
Detailed Comments
—————————————————————————

ome further clarification would be advisable if the paper is finally accepted.
In particular, the author(s) should clarify either a) a theoretical framework that they’re using to analyze this issue, or b) further clarify the experiments and usability studies that would be needed to test CAS. I’m not entirely sure that this is a research paper.

============================================================================
REVIEWER #5
============================================================================

Reviewer’s Scores
———————————–

Relevance to Information Architecture: 9
Contribution to Information Architecture: 8
Originality and creativity of ideas: 8
Quality and rigor of research or design approach (for research posters): 7
Clarity and organizatrion of presentation and writing: 7
Potential of Topic for Engaging Audience: 8
Potential to Aadvance the state of the art: 7
Significance of findings: 6
Overall Rating: 7
Recommend: marginal

—————————————————————————
Detailed Comments
—————————————————————————

The CAS approach is promising. I would be most interested in seeing actual experimental work, as opposed proposed experiments. At this stage there is still a need for papers that articulate important and promising research directions. But any experimental work would make this stronger and help illuminate the possibilities.

Dear Mr. Jason Toal:
On behalf of the IA Summit 2006 Program Committee, I am delighted to inform you that the following submission has been accepted to appear at the conference:
The Life of Tags
The Program Committee worked very hard to thoroughly review all the submitted papers. Please repay their efforts, by following their suggestions when you revise your paper.

I guess we should actually write the paper now

Tagging as an interface trend in current social software architecture relies on user input to assign meaning to a given object, typically a link, image, or document. This process has given rise to a term called ‘folksonomy’ or ‘tagsonomy’, a play on words from its more formal predecessor, ‘taxonomy’. Unlike its namesake however, folksonomy is predicated on the idea that social intervention can give rise to a more meaningful web experience. As the internet continues to grow at a steady rate, making things more findable and useful is becoming increasingly difficult. In opposition to this, there is a continued pursuit to expand the underlying structure of the web through an ontological framework that will form a stable basis of meaning. It is felt that a massive all encompassing ontology that users can rely on to extract meaning from the web, and at the same time is governed by rules that software can interpret will be the best environment to make sense of this expanding space.

We see the need for a combination of machine readable structure, or rules, upon which socially driven input can exist upon or within. One of the greatest problems with folksonomy is the variation from which a user can assign meaning or tag an object. Semantic differences alone account for much of the disparity in meaning, but even more unknowable factors such as using a plural form of a word can cause a great separation of ideas that should otherwise be grouped together. Also, there is the capability to dilute the meaning of a tag through inconsistent or malicious use, as would be the case of spammers. Ontology and other hierarchical based approaches are too fixed and do not take into account, or make possible an evolving dynamic space that reflects more the natural world we exist within.

A Cellular Automata System (CAS) is a dynamic system that arranges itself based on the state of the system and the state of the elements within the system. Each element in the system follows a simple set of rules that define how they communicate with each other, and how they respond when they encounter another element. By following the rules in aggregate an order can emerge within the system that is not apparent within the rules themselves.

Both CAS and tagsonomy are self-organizing systems that rely on the isolated acts of the individual to create a higher-level order. However, the organization in the current crop of tagsonomy systems (del.icio.us, flickr) is very low level when compared to other bottom up systems like CA. The difference being individuals in a CAS follow universal rules as they move through the system, while taggers do not. It is the rules that make (CAS) work.

Our paper will pursue potential opportunities for adding a usable layer of rules to the process of tagging. We will discuss the potential of ‘adding metadata to the metadata’ through weighting of tags or sequencing of tags through more formal structures. Where possible, we will propose experiments and usability studies that could test the validity of our theories.