Hacker News Reader

The details at http://en.wikipedia.org/wiki/Turnitin may be of thin help. At least one Canadian student has protested, and won. The number one point I took from it would be to advise her to not agree to the clickwrap license. Then she could at least protest about the abuse of her copyright.

If you two want to really stir things up, you can start a company which does text analysis of student essays and hire her with an exclusive contract to provide corpus material.

Some background for the company. "katengine.com" appears to be owned by Knowledge Analysis Technologies. http://www.forbes.com/forbes/1998/1228/6214047a_print.html has more info about the company, which was acquired in 2004 by Pearson Education http://kt.pearsonassessments.com/whyChooseUs.php .

If the teacher will be using the 'Intelligent Essay Assessor' from http://www.mywritinglab.com/learn-about/writing-practice.htm... then yes, it's a variation of LSA, as you found. There's a fact sheet there with a picture of what the GUI might look like.

Based on that name, http://www.d.umn.edu/~tpederse/Courses/CS8761-FALL04/Project... says "This was developed in the late nineties by Tom Landauer and his team and is based on Latent Semantic Analysis. In this, using a matrix transformation technique known as Singular Value Decomposition (SVD), deeper hidden relationships between words and documents are discovered, and these relations are used to understand relation between words and sentences."

I found http://writing.ucdavis.edu/news-1/carl-whithaus-essay-publis... which describes an essay critical about machine grading. I couldn't find that essay though.

http://www.huffingtonpost.com/2012/03/29/robo-readers-the-ne... quotes: "A prankster could outwit many scoring programs by jumbling key phrases in a nonsensical order. An essay about Christopher Columbus might ramble on about Queen Isabella sailing with 1492 soldiers to the Island of Ferdinand -- and still be rated as solidly on topic, Shermis said." as well as "Computers also have a hard time dealing with experimental prose." So she might gum up the works by using an unusual but legitimate style.

Speaking of gumming up the works, I suppose that the essay must be submitted to the teacher in a text or word processor format? That is, hand-written or printed essays are prohibited? Or embedding a strange font which shows a "b" for "a", etc., and where the text is transposed accordingly? Pull quotes to throw off the formatting? Hidden text, colored white? That would be subterfuge.

Edit: Pearson also has "second-tier graders", says http://news.slashdot.org/comments.pl?sid=2757605&cid=395... , with a human involved.

"A prankster could outwit many scoring programs by jumbling key phrases in a nonsensical order. An essay about Christopher Columbus might ramble on about Queen Isabella sailing with 1492 soldiers to the Island of Ferdinand -- and still be rated as solidly on topic, Shermis said."

To be smart enough to prank on a subject and pass would mean a pretty in depth knowledge hence the mark would be correct.

In fact a human would incorrectly say the student was bad at the subject so it makes the computer superior in marking students actual knowledge of the subject :)

I don't agree. It seems to imply that a way around robo-graders is to list facts, and don't worry so much if the essay hangs together. In any case, your view and mine are hypothetical. I haven't seen an analysis of how well the system works in practice, or how easy it is to spoof. And it looks like it's the system at a whole which must be analyzed, if there's a second tier human layer which might review odd-ball papers.

Assuming there are existing lecture notes highlighting key information, you could probably pump them into the sort of "article spinners" used by SEO spammers and produce something that looked like accurate, original writing to a computer algorithm, without having any actual knowledge of the subject at all.

Studying history (even at junior high school level in the UK) is more about interpreting sources and making reasoned arguments, than simply regurgitating facts and references. You can parrot or parody a list of facts and references you've been given whilst being a hopeless history student. Sounds like a machine comparing with a "model essay", meanwhile, would penalise those students foolish enough to construct original arguments and support them with reference material beyond the compulsory reading list.

I'm not sure what the issue is here. I bet checking and scoring based on a combination of grammar, spelling, and a rough relationship to the topic at hand would produce pretty good results. Further, if you feel you were graded unfairly, the professor specifically said he will review it personally.

He's probably already overworked and underpaid. Why not help him and the college implement technology that could prove to be very helpful.

Only if students are also allowed to submit essays generated by a computer.

This is an interesting idea. If you had access to the grading algorithm's source code, wouldn't it be easy to randomly generate essays that always score 100?

I don't see why. The code could be very complicated to understand, or the grading mechanism could be hard or impossible to reverse. (Just like cryptographic hashes.)

Better would be access to the training corpus. Then use that to train, say, a Markov generator, and have it replace words from a thesaurus, in hopes of bypassing a similarity detector.

It would be fun to try it with SCIgen (http://pdos.csail.mit.edu/scigen/)!

So I guess I think that you might be well advised to go to the dean and say "listen, we pay you good money, and we expect our instructors to actually do their jobs, rather than to farm them out to machines."

At the same time - and I mean no disrespect - this is community college we're talking about. The purpose of community college is to pick up skills, rather than a liberal arts education; you should, at that price, expect to get a somewhat reduced response.

I am with you here; this seems lazy and inappropriate.

But you get what you pay for. OKCCC is $75 a credit hour. By contrast, my alma mater currently charges $24,000 per year (out of state,) and nominal is 15 credits a year. Even trade schools tend to cost several hundred dollars a credit hour.

At that kind of price difference, you have to realize that you're forcing a lot of corner cutting.

You could also argue that the state and community pay a lot towards this professor and, depending upon active participation, his administration. He or they appear to be "ripping off" these entities.

Communities fund community colleges particularly for the purpose of building localized skill sets, thereby improving said communities.

There's a lot of financial stress in the U.S., these days. Nonetheless, the community should know what they are getting. Both for their money, and in terms of opportunity when it is their turn (whether directly, or through the workers they hire).

Many community colleges (including some in my area) used to have quite good reputations. It was not simply a linear function of "the student gets what the student directly pays for". Nor, I think, should it be. There should be some accommodation for merit, different rates of "maturity" or readiness and timing for college academics, hardship, etc. Also for opportunities for life-long education. Community colleges play an essential role in fulfilling this. And it shouldn't be a cut-rate, McEducation.

You are inappropriately harsh. Students go to a CC for several reasons. Some go to a CC because it's cheaper and closer to home. These people plan to transfer to a 4 year college after getting an associate's degree. Others didn't do so well in high school and want additional remedial courses before going to a 4 year college. In many states, it's easy to transfer from a CC to a 4 year state college, because the additional work needed for an AA degree is more meaningful than the work to graduate from high school.

For example, "The regents of the University of California report that 30 percent of UC graduates attended a community college before transferring to the UC."

Yes, as you say, CC is also for career education. It's also for non-credit/continuing education. I took several CC classes in the latter category.

And the comparison of OKCCC to your alma mater is almost useless. Better would be to compare to Oklahoma State where it's $147.50/credit hour. (There's another roughly $110/credit hour in "fees". For an arts&sciences degree, the overall cost is about $275/credit hour. For OKCCC it's $99/credit hour, including fees.)

The "corner cutting" doesn't need to be in the education. The mandatory fees for OSU include "student union renovation", "campus recreation", "Academic Excellence", "Library Automation & Technology", "Life Safety & Security", and others. Since many people go to a CC because it's local, they are less likely to spend time in the student union or in on-campus recreation so the corresponding fees are smaller. The library at a CC is likely smaller, and less in need of automation.

Also, the lab equipment for a two year program doesn't need to be as advanced. My 4 year college upper-level physics lab had microwave generators, an x-ray crystallography machine, and a radiation source. Why should a CC also have them?

These aren't cutting corners, these are inappropriate expenses for the goal of the students and the school.

That said, a CC program isn't going to have tenured research professors in a field teaching an introduction course, and won't have the fellowship/intern programs in place, or foreign exchange programs. These can (indirectly) affect the quality of education.

But to say that these are all cut corners is to idealize what an education must be.

> Students go to a CC for several reasons.

Yes. For some time I was one of them. I am uncertain why you interpret what I said to be harsh.

Respectfully, it seems to me like saying "look, at the price you pay at McDonald's, there's not enough money floating around to get a filet."

This is realism, not harshness. It costs a lot of money to run a school, and the US doesn't really fund its community colleges.

.

> And the comparison of OKCCC to your alma mater is almost useless.

It would be nice if, while saying I'm being inappropriately harsh, you differentiate between something that's useless and something you merely do not understand the use of.

.

> These aren't cutting corners, these are inappropriate expenses

Sure. They're also not even slightly the dominant costs in running a school, and not what I referred to.

But you're right, the topics you raised, that I did not, aren't particularly cut corners.

.

> That said, a CC program isn't going to have tenured research professors in a field teaching an introduction course, and won't have the fellowship/intern programs in place, or foreign exchange programs. These can (indirectly) affect the quality of education.

My my. And here I thought you didn't agree with me.

Of course, these also aren't the dominant costs in running a school, and also aren't what I was talking about.

As a point of preference, I generally ask someone to explain themselves before calling them wrong in public. But maybe that's just because I don't like arguing with my own guesses about what someone meant. YMMV.

.

> But to say that these are all cut corners is to idealize what an education must be.

I would request that you re-read what I said more carefully. The things you point out aren't actually related to what I was discussing.

College professors grading essays purely by computer - is this legit?