This semester we examined two common techniques for representing, organizing, retrieving, and filtering information:

• the deductive Boolean logic of classes and their combinations, and
• the inductive Bayesian logic of probabilities conditioned on past events.

As Bernhard Rieder observed, each of these techniques “implies a particular way of doing things” but can “be applied to a wide array of domains.”

For example, Boolean logic can be applied to classify and filter scientific articles, works of choreography, consumer products, or potential romantic partners, if one can come up with a way to represent these things as belonging to classes defined by necessary and sufficient properties.

Likewise, Bayesian logic can be applied to these same domains and many others, if one can represent these things as having quantified properties suitable for training a statistical model. In either case, for any particular domain of application, there are “many decisions to be made concerning the units to take into account, the parameters to specify, the tweaks to apply, the outputs to produce, and so forth” (Rieder).

For this question, you will identify and describe two different applications of information classification and filtering. The first should be an application for which you believe Boolean logic to be the preferable technique, and the second should be an application for which you believe Bayesian logic to be the preferable technique.

For each of the two applications that you identify, you should:

• Describe the application briefly but with sufficient detail to provide context for your argument in favor of the chosen technique. Who uses it, and why? How do they use it? What is the “stuff ” being classified and filtered? How are the outputs of classification and
• Explain how the technique you’ve chosen could be applied. If the application is the one for which you believe Boolean logic would be preferable, explain how the “stuff ” could be represented as classes defined by necessary and sufficient properties, and how someone using the application would use these classes to classify and filter. If the application is the one for which you believe Bayesian logic would be preferable, explain how the “stuff ” could be represented as things having quantified properties, and how these representations could be used to train a statistical model, and how the estimated
• Give two reasons why your chosen technique is preferable for this particular application. Again, do not just identify general strengths of the technique—instead give specific reasons why the technique might be preferable for this particular application.

filtering presented or used? The application may be purely hypothetical, or it may be based on a real application that you are aware of. See the next page for an example.

probabilities produced by the model could be used to classify and filter. Do not just give a

general explanation of how the technique works—instead focus on some of the specific decisions to be made for this particular application.

4.Finally, identify one potential problem with applying the technique in this way. What limitations might the technique impose, or what negative consequences might result? One more time: do not just identify general limitations or drawbacks of the technique—instead describe a specific negative outcome that could result from using this technique for this particular application.

Example of a description of an application, like you might write for part 1 above (do not

use this example in your own answer):

The application is intended to help faculty at a large public university identify students who might struggle in their courses. The things being classified and filtered are students, as represented by their transcripts and other aspects of their student records. Specifically, students are classified as prepared or unprepared for particular courses. Faculty see these classifications on their course rosters. They can use this information to decide who will be allowed to register for the course, or to determine which students may need extra attention during the semester.