Create a new Assignment Submission Page titled: "A6 - Your Name".  Make sure to tag it with the correct assignment tag ("A1" or "A2", etc). You must do this to ensure that we can see your assignment once you submit it. If you fail to do this or forget to tag your assignment, you may receive a late penalty since we will not be able to find your work.

Assignment 6: Personal Information Management and Distributed Classification

Posted: October 29, 2010

Part 1 Due: November 3, 2010

Part 2 Due: November 10, 2010

Author:
Bob Glushko,
glushko@ischool.berkeley.edu

Lead TAs: Kimra McPherson, kimra@ischool.berkeley.edu (part 1), Jess Hemerly , jhemerly@ischool.berkeley.edu (part 2)

Course: Information Organization and Retrieval (INFO 202)


Assignment Overview

In this assignment, you will:

1. Come up with 10 tags for a photo

2. Cluster a large list of tags into semantically equivalent concepts

3. Develop three principles you would use for tagging, if you were in charge of Flickr

4. Reflect on your experiences

Deadline

You must submit your work by creating a new assignment submission page before 9 a.m. on Wednesday, November 3, 2010 for part 1 and before 9 a.m. on Wednesday, November 10, 2010 for part 2.

Submission Requirements

For part 1, you will submit a text file called YourNameA6tags.doc. For part 2, you will submit a text file called YourNameA6report.doc

Detailed Instructions

Part 1: Tag the Photo

Start out by taking a careful look at the photo attached to this assignment page. 

Individually, make a list of 10 tags that you would assign to the photo and save them in your YourNameA6tags document. Don’t worry if all of your tags don’t follow the description rules we’ve discussed in 202. You should strive for at least half your tags to be ones you’d consider “good” – but part 2 will be more fun if some of the tags are a little odd. You’ll see why in a second.

Part 2A: Cluster the Tags

On Wednesday, November 3, the TAs will send a message to your section e-mail lists with a link to the set of tags you should use for this assignment. The list will consist of all the tags generated by one of the other sections – plus perhaps a few others.

Take those tags and cluster them into semantically equivalent sets. Think back to what you’ve learned about description and categorization and see where you can group conceptually similar words together. (As an example, perhaps you’d want to cluster “tree” and “trees” together — but maybe “elm” is a separate concept.)

In your YourNameA6 report file, arrange the tags into your clusters. Then decide on a name or “authoritative tag” for each cluster. For example, one of your clusters might look like this:

Tree

Tree

Trees

You may have outlier tags that don’t seem to fit into any cluster. That’s OK – but make a note of them anyway. You’ll be talking about these outliers in your reflection.

Part 2B: Develop Tagging Principles

Now that you’ve seen “the wisdom of crowds” in action and had to deal with the variety of ways people tag the same photo, we want you to develop some principles for useful tagging. Specifically, we want you to imagine that you’re in charge of developing the tagging interface for some new photo-sharing website. How would you guide users to tag in a certain way?

In your YourNameA6report document, create a list of three objectives you’d want your system to achieve. For each objective, list at least one guideline that users would follow to meet this objective and explain why you think it would work.

For example, something like “Reduce multiple forms of the same basic tag” is an objective, and something like “Auto-complete a tag once someone has started typing” could be a principle. By “explain” we mean that you should relate something you’ve learned about information organization in expressing why you believe your principle will help users meet your objective.

Part 2C: Reflection (or: So, Are Crowds Wise?)

In your YourNameA6report document, reflect on your experiences tagging, clustering tags, and developing principles for tagging. Specifically, you should answer the following questions:

- How easy or difficult was it for you to develop your personal list of 10 tags?

- How easy or difficult was the process of clustering the tags? Did your clusters encompass all the concepts in the tag list, or could some concepts not be clustered effectively?

- Were you surprised at the level of tag convergence on the list you were given?

- Overall, what did you think of the tag list you received? Were there any oddities?

- Overall, what do you imagine would be the benefits of implementing tagging principles like those  you suggested in part 2B? Would there be drawbacks?