Create a new Assignment Submission Page titled: "A5 - 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 5: Animal Kingdom Ontology

Posted: October 11, 2010

Due: October 18, 2010

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

Lead TA: Kimra McPherson, kimra@ischool.berkeley.edu

Course: Information Organization and Retrieval (INFO 202)


Assignment Overview

In this assignment, you will:

1. Define equivalence classes for 15 animal instances.

2. Sort those classes into a hierarchy of animal types.

3. Create a diagram of your ontology.

4. Write definitions for each part of your ontology using hypernyms and hyponyms. 

5. Reflect on your experience.

Deadline

You must submit your work by creating a new assignment submission page before 9 a.m. on Monday, October 18, 2010. Late assignments will not be accepted unless you have an exceptionally good excuse.

Submission Requirements

You will submit a total of 3 files (zipped). Name your zipped file YourNameA5.zip. The first file will be a spreadsheet (modeled on the sample spreadsheet attached here) called YourNameA5.xls. The second file will be your diagram, which will be called YourNameA5.pdf. The third will be a brief reflection (YourNameA5Reflection.doc). Regardless of what method you use to make your diagram, please save a copy as a pdf so we can be sure we’re seeing what you intended for us to see.

Detailed Instructions

In this assignment, you’ll be returning to your “ark” from A4: the set of 10 original animal instances plus whichever bonus set you selected. This time, you’ll be developing a hierarchical classification scheme.

The goal of this assignment is to give you more practice thinking about categories and category membership, abstraction, classification, and ontology. You’ll also learn a technique for naming and describing a system of categories so that you can clearly convey their meaning to others.

Part 1: Identify your types/equivalence classes

Round up your animals by returning to your 15 instances from A4.  In a spreadsheet, following the format of the attached sample, create a list of these animals. (Hint: Remember that each of the instances came with a name attached for A4. You’ll probably want to continue using these names.)

In this part of the assignment, you’re going to start generalizing away from the specific instance you were given. For each animal instance, identify a “type” or equivalence class to which that instance belongs. For example, if we were classifying musical instruments, and you’d been given a picture of a drum set, you might pick something like “rhythm instrument” as the type. Or you might choose something more granular or more abstract than that — remember, as always, you’re making a choice about the level of abstraction you use.

 One thing you don’t want to do here is make your equivalence classes so specific that they can’t describe anything but the instance you were given. (Making your equivalence class “yaks” for an instance called “yak” seems awfully convenient, but it’s a little too easy.) You should be able to think of some common features held by all members of your equivalence class, as well as some other instances that would bit into the class. 

As you’re making your first pass through the instances, do not stress out too much about naming these types. You’re likely to go back to them and revise them as you progress through the assignment. If it’s starting to make you feel crazy, our advice is to come up with something temporary and move on; new ideas might pop up once you’ve started to arrange your hierarchy. 

Part 2. Organize your types into a hierarchy

Now that you’ve taken a crack at identifying types for each of your animals, begin arranging them into a hierarchy. The top or root element of your hierarchy will be “animals.” The bottom level of your hierarchy will be your instances. When you created your types in Part 1, you added a second level to the hierarchy — more abstract than your instances but less abstract than “animals.” What you’re doing now is adding one more level of abstraction — a new level between your types and “animal.” These are hypernyms or "super-types."

Think of this as a sorting task. (Sometimes it’s even helpful to write your type names down on pieces of paper or sticky notes and physically sort them.) As you sort, you may discover that some of your original types are too narrow. You may also realize that they’re too broad and don’t leave you enough room to insert another level before getting to “animal.” That’s OK! Revise your type names as many times as you need to and record them in your spreadsheet.

At this phase of the assignment, it’s important that you strive for a consistent level of abstraction among your “super-types.” Again, try to think of some common characteristics that would be shared by all members of that super-type. You won’t be as precise as the actual biological classification of the animal kingdom (species -> genus -> order, etc.) -- and you certainly don't need to follow the biological rules to a T! -- but keep that structure in mind as you look at your own hierarchy. For example, if we had musical instruments as our root element and our next level down included both “clarinets” and “stringed instruments,” that might be a sign that the classification wasn’t maintaining a consistent level of abstraction.

Part 3: Create a diagram or visualization of your hierarchy.

This does not have to be fancy. We repeat: This does not have to be fancy. Start with “animal” at the top or root of your visualization, then your “super-types,” then your types. You can use any tool you wish (including drawing by hand and scanning your drawing), as long as it allows you to represent the hierarchical relations in your ontology. 

Part 4: Define your types and “super-types.”

Now, you’re going to write a definition for your equivalence classes and your “super-types” such that an ordinary person would be able to categorize new instances. You’ll be following this formula for definitions:

Hyponym = {adjective+} hypernym {distinguishing clause}

Example:

Say I’m classifying instruments. We’ll take “Instruments” as my root element.

My first instance was a bass clarinet, and I assigned it to the equivalence class called “clarinets.” Then, as I created my ontology, I sorted clarinets, saxophones, and flutes together into a “super-type” called “woodwinds.”

My definitions might then look something like this:

Clarinets = {reeded} Woodwinds {that are approximately cylindrical in shape and have numerous keys}

Woodwinds = {reed or flute} Instruments {that produce sound when air is blown into them}

 Remember that your definitions should reflect things that are true for all members of a type. A good sanity check at this stage is to make sure you can think of a hypothetical second instance for each type.

Record each definition in your spreadsheet.

Part 5: Reflect on your experience

In your YourNameA5Reflection document, write a paragraph or two about the approaches you used to identify equivalence classes and organize them into “super-types.”

Some questions to guide your reflection: What was your thought process like? What specific tips from the readings or lectures did you draw on? Were there any “outliers” that you had to work especially hard to fit in? Were you able to keep your "super-types" to a consistent level of abstraction, and how did you do so? Was this harder or easier for you than the faceted classification of A4, and why?