Sunday, November 30, 2008

Poorly Structured Exam Questions

For engineering exams, I absolutely hate poorly structured exam questions, particularly questions that have a heavy weight. It is my opinion that such questions serve little purpose.

Before I proceed to expound upon my views, allow me to more clearly describe the item of my ire. Poorly structured exam questions are exam questions that only contain 1 main question (no sub-items), and yet have a considerably high weight assigned to it. For example, "Describe how XYZ can be ZZZ. [25 marks]" is one such problem. Such questions require the student to perform certain tasks, but do not clarify what exactly is required of the student.

Now, imagine a hapless student attempting to address such a question. Given such an ambiguous problem, what can the student do? He has no choice but to adopt the safest approach, which is to construct an answer that is as foolproof as possible. The result is a voluminous answer, which is comprehensive but also largely irrelevant.

It puzzles me to learn that markers of exam scripts often have reference marking schemes, which detail which points are correct, and of the proper score to assign to those points. If there is already a set of criteria which the examiners are looking for, why should the criteria be kept secret? What harm would it pose to give students an idea of what answers are required?

It would be infinitely easier to properly structure the question into smaller and more detailed parts, or to provide a clear description of which points or areas should be discussed in the answer. Adopting such actions would clearly reduce the amount of irrelevant material in the answers. It would also be easier for the markers to mark the exam scripts. A mutually beneficial scenario, in other words.

It is my opinion that poorly structured exam questions are the mark of a lazy problem setter. However, such laziness does not go unpunished, since the same setters would have to suffer the large volume of inflated answers that is a direct consequence of their laziness.

Secret Reseach Project 3: Authorship Attributor

During the term, I developed a program that was able to attribute a document to its correct author. In other words, given a sample document, the trained program would be able to indicate who wrote it.

The theoretical basis for the authorship attributor is that different authors have different active vocabularies and different word preferences. Hence, this difference can be exploited to differentiate between writings of different authors, as long as enough of their prior work is known.

The program employed genetic algorithms to evolve simple rule-based classifier systems. The final performance on the test set, the Federalist Papers, was generally good, although slightly inferior to existing methods.


























































P L M C Average % Correctly
Classified
Average No. of
Active Rules
100 25 100 5 91.67 4
50 25 100 5 78.33 5
100 10 100 5 88.33 2
100 25 10 5 73.33 11.2
100 25 100 25 83.33 4.6
100 25 10 25 78.33 7.6
Table: Performance of Classifiers for Given GA Parameters

The key weakness of my proposed approach is that is too simple, employing only a voting framework of rules. Furthermore, the individual rules consider only the relative frequencies of pairs of word, which is an approach that is not generalizable. In other words, while the existing approach may be useful for a pairwise classifier, it is likely to be useless for developing a universal attributor capable of differentiating between any number of candidate authors.

Still, in terms of potential, I believe that this particular research project has more potential for improvement than my previous research projects. In particular, the rule combining system can be improved from the voting framework currently used. Furthermore, the rules used can be made more general versions with little modification. Additional criteria could also be included.

I hope to revisit this problem at a later point in time.

Saturday, November 29, 2008

A Physics Thought Experiment

Consider the scenario illustrated by the following diagram:


There is a hot object initially at temperature T. The hot object is instantly teleported into a larger container, which contains a perfect vacuum and has a perfectly reflective inner surface. The closed container is perfectly reflective at all frequencies.

Assume that the container is magically suspended in the larger container. As the container contains a perfect vacuum, no heat is lost by conduction or convection. Heat can hence only be lost by radiation.

Does the final temperature of the hot object depend on the size of the larger container?

Wednesday, November 12, 2008

Factoring on the Go

Yesterday, I started factoring numbers. Apart from being a method to make waiting less mundane, it also works the brain.

I managed to factor the numbers from 1 to 100 while waiting for my bus to arrive. This would have been a minor feat, were it not for the absurd frequency of the bus service.

I also managed to factor 638 before 6:39pm. After more practice, I hope to be able to perform faster factoring.