October 2006 Archives

So, for school I'm doing a natural language processing project in which I develop an application to look at determine blogs and determine, based on the blog content, certain features about the authors. If any fellow bloggers are interested in participating (and I would be greatly indebted to you if you decided to participate), please fill out the questionnaire below and email it to me by October 31, 2006 (preferably sooner).

1) What is your full name?

2) What is your email (I will keep this private)?

3) What is the name of your blog?

4) What is the URL of your blog?

5) What is your date of birth?

6) Are you Male or Female?

7) Would you characterize yourself politically as a Democrat (Liberal), Republican (Conservative), Independent, or Other?

7a) If Other, would you please provide details?

8) Are you the only author of your blog?

8a) If not, are you the primary author? About what percent do you contribute? What is the makeup of your blog’s other authors?

A slightly more detailed explanation of the project can be found here. A more thorough, technical treatment can be found here. For those who still have questions, you can email me.

Thanks very much to those who decide to participate. And, if you know anyone else who might be interested in participating, please forward them this blog, or my email (John.Umbaugh@gmail.com). Thanks, and have a good day.

Oh man, that 543 homework killed me. I submitted it at exactly midnight - fifteen minutes before the due datetime. I did in fact get it working. The second problem - worth sixty percent - was the worst; to compound it enormously, I carelessly read the directions wrong and wasted a lot of time. Luckily, the third and last problem was no big deal comparatively. Only took a couple of hours.

Each homework is worth twelve percent of our class grade. I went in to talk to the professor today, and it seems like a lot of the other students were also having problems. It seems that at least a couple of them were quite a bit further behind than me. That was strangely comforting... It made me feel like I wasn't an idiot.

I have to admit that the project - though stressful - was a ton of fun. I'm glad that I got to do it, and I really did learn a lot. And I'm becoming a Lisp master! Well, not a master, but at least I can speak it now. Got most of the grammar down.

So, over the years I had come under the impression that, as I've aged, I have somehow grown cooler. A video that my buddy Bowen sent me proved to me that, in fact, no. He wrote something along the lines of, "I saw this and I thought of you [guys]." Bit of a blow to the ego. Please enjoy Weird Al Yankovic's hilarious video White and Nerdy.

Yeah, I fit uncomfortably many of those stereotypes. I think that those hoodies are for sale somewhere on the web... I wonder what my students would say if I showed up in class wearing that?

Hoo boy... this EECS 543 homework has me down. It's due a little over twenty-four hours from now, and I'm nowhere near being finished. I do think, however, I've got a handle on the problem that's been dogging me. After like ten hours...

So, last Thursday, right before I went to the curry shop and before my NLP class, a freckle-faced but sunglass-ed guy wearing a Bob Marley t-shirt outside of West Hall was accosting passersby with, "Excuse me, sir, do you have time for the environment?" Joe Cool. I thought I could slip by him - you know, hide amongst the moving crowd like a fish within its school - but I somehow failed. "Hello, do you have time for the environment?" he said, brandishing a flyer, at a distance that seemed strangely far, at least under normal conversational conventions. Was he talking to me? Well, I made eye contact - or at least I think I did; tough to say with those sunglasses. It would be rude to say nothing, but I had to get lunch and get up to class, pronto, so I politely declined. "Oh, OK, well have a good day," he said in a sing-song voice.

I felt awful. Here was a guy who was sacrificing his time for a cause in which he believed - in which I believe - and I callously breeze by him. All for the sake of curry. And, he was so genuine, that he actually wished me pleasantries after my blow-off. I must not care about the environment as much as I should, I thought to myself. What kind of heartless jerk doesn't care about the environment? I must be a terrible person.

Then, I suddenly stopped feeling awful, and started feeling annoyed. I realized that, in fact, the whole enterprise had been varefully engineered for success with either outcome. If the person decided to engage the caller, that person became more of an ally against environmental destruction. On the other hand, if the person decided that he or she were too busy, the caller would - seemingly innocently - make the person feel guilty about his or her callousness, and cause her to engage towards environmental idealism at a later date - perhaps with even greater vigor. Now, I care about the environment as much as the next guy, but I detest these kinds of social manipulations. Next time, I decided, I would fight back.

Well, next time came a lot earlier than I expected. Shortly after my NLP class ended, on my ten minute walk to the bus stop, I was again accosted by a young man wearing a tan cowboy hat, sheaf of flyers in hand. I got a better look at them this time; it looked like Al Gore's - or somebody who resembled Al Gore's - face figured prominently in high-contrast black and white. "Sir, do you have time for the environment?"

Without saying anything, I approached him a step or three. A couple of seconds elasped. I looked him squarely in the eye, and, without a hint of mirth, asked him, "pro or anti?"

"Wh- wha-... what do you- Pro! Of course!" He lost his cool a little bit there.

I smiled and politely declined. "Uhh... OK then, have a nice, uhhh, day...." I heard him say as I started to walk away.

Childishly, I had a new spring in my step! I walked away, strangely, somewhat Satanically satisfied with the notion that I might have implanted in him the idea that there were actually - unsettlingly - people who actively wished harm to the environment. Oh, when will I grow up?

Instead of reading the books that I should be reading - that is, for my Natural Language Processing classes - I've been working my way thrgoough a book I heard about on the Diane Rehm show. The interviewee on this particular episode was Edward O. Wilson - naturalist, entomologist, myrmecologist, and writer. Wilson's and co-author Bert Holldobler's book Journey to the Ants: A Story of Scientific Exploration is a really incredible and eye-opening introduction to myrmecology - that is, the study of ants.

Here's a fun fact about ant anatomy: many species have what is called an organ called a crop - or, as Wilson and Holldobler frame it, a social stomach. From the book:

...the worker on the left induces regurgitation by touching her forelegs to the donor's head.... the donor, on the right, passes liquid from her crop the storage organ that serves as the "social stomach," through her esophagus into the mouth and crop of the recipient. Small amounts of the food are also passed from the crop into the midgut to serve as nourishment for the donor. Waste material is passed out through the rectal bladder.

The chapter about ant communication mechanisms - mainly pheremonal, but also through vibrations, through tapping, dancing, and even some visual cues. Pretty incredible.

Currently I'm reading a chapter entitled "War and Foreign Policy." Listen to this:

The honeypot ants, skilled communicators in a violent world, also maintain a long-term balance of power with minimum loss of blood - or (to use the correct entomological term) loss of hemolymph. Certain colonies, especially those with the greatest number of large-sized workers, the major caste, dominate neighbors, bluffing them away from the tournamament arenas and into smaller foraging zones closer to their nests.

Somehow the individual combatants are able to assess the strengths of the enemy and use that knowledge to respond by choosing either greater boldness or deference. How can a simple insect make such an evaluation, and arrive at such a decision? It was obvious to Holldobler that the ants can never get a bird's-eye view of the whole arena. They are unable to count the numbers of combatants from the two colonies. Not even the entomologist can accomplish that, at least not without using stop-action film analysis conducted across the whole course of the engagement. Holldobler proceeded to do just that, tracking scores of honeypot-ant contests in the desert. He took the data to the theoretical biologist Charles Lumsden, and together they studied the problem. Eventually they came to realize that there are at least three ways the any workers might indirectly assess enemy strength. They can "count heads" while shifting from one combatant to another. If their nestmates outnumber the enemy - say three to one - they will be subjectively aware of the imbalance in their favor and more inclined to press forward. If the reverse, they will retreat.

Swarm intelligence in social insects. Swarm intelligence is everywhere.

Heretofore, on Tuesdays and Thursdays before my Natural Language Processing class on the central campus at UMich, I haven't ventured beyond the culinary confines of a semi-good Mediterranean restaurant, or - more often - a chili dog from a mediocre hot dog stand. But no more. Across the street the other day, I discovered an Indian curry stand. A stand.

Today I bought lunch from it and took it up to the classroom. Only $4.75 for one bowl of a - well, a type of dal curry - and one bowl of rice. Not too spicy, and didn't make me feel ill. Tasted pretty good, actually. Definitely not the best curry I've had, but pretty darn good nonetheless. I'll be eating there as long as they're there (that's a lot of ðÆrz). I wonder how late in the season they'll hang around?

Ann Arbor has some of the best places to eat imaginable.

I've been reading a lot about collocations in my EECS 597: Language and Information course at the University of Michigan. There's an interesting passage on page 152-3 of the textbook, Foundations of Statistical Natural Language Processing, that I wanted to share with you.

A good example of the type of problem that is seen as important in this contextual view is Halliday's example of strong vs. powerful tea (Halliday 1966: 150). It is a convention in English to talk about strong tea, not powerful tea, although any speaker of English would also understand the latter unconventional expression. Arguably, there are no interesting structural properties of English that can be gleaned from this contrast. However, the contrast may tell us something interesting about attitudes towards different types of substances in our culture (why do we use powerful for drugs like heroin, but not for cigarettes, tea, and coffee?) and it is obviously important to teach this contrast to students who want to learn idiomatically correct English....

The top five two-word collocations beginning with strongwere as follows: support (50), safety (22), sales (21), opposition (19), and showing and sense, tied with (18).

For powerful, the top collocations were force (13), computers (10), position (8), men (8), computer (8). This is interesting, since powerful and strong, as units, could generally be regarded as very close synonyms.

Also interesting was the presence of men as tied for fourth, and man as the sixth. For collocation 1-component powerful, man occurs as the twentieth most frequent 2-component. Nowhere is woman or women listed, however. Women, our culture seems to reiterate over and over again, are neither strong nor powerful. That is strictly the domain of men. Well, and Oprah (joke). And this corpus is from New York Times newswire. I wonder what other beliefs our culture is propagating with its collocations...

Furthermore, I wonder if it would yield any fruit to look at the ratio of these kinds of collocations against corpora in different genres (or even languages) in order to get some measure of comparative "enlightenment" (semi-joke).

I've been pretty busy lately with my GSI position. The students have their first major project due tomorrow, and I think it was a bit difficult for them. But not too difficult. The project was to create a simple arcade game - in three weeks. Some of the projects that I've seen have been really great. I'm excited to demo the games on Friday; I think the students are, too!

Last week was a good one in EECS 494. We had three guest speakers come in, and they all were interesting. Michael Malone from EA and Alex Kerfoot from Maxis (and a UMich graduate) were both particularly good. I was halfway inclined to get back into the games industry after hearing their talks!

This week, I've been preparing workshops for the interactive fiction toolkits Adrift and Inform. Both are applications and sandboxes for developing Interactive Fiction - that is, for example, text adventures in the tradition of Zork and Adventure. And, wow, are these tools sophisticated.

Adrift is comparatively easy to learn. First, there's no programming - it's all GUI-based. And it's got a very intuitive interface. There are very useful features such as an auto-generated map, dependency graphs, and a browser (which, unfortunately, I can't get working in the demo version). One feature that has really gotten me excited is the ability to conduct conversations with players in the game. It's still very simplistic, but with a little bit of tweaking, I think you could get some very believable behavior there. Maybe enough to pass the Turing Test. In a game setting, anyway.

Before class I created a kind of mini-tutorial screencast for Adrift with CamStudio. I say mini, but it turned out that the generated .AVI was 1.3 Gigabytes, and 50 minutes long (currently I'm trying to convert it to .ASF to see if it gets any smaller. Also, the prof can't hear the audio on his Mac...). In class, we had a really good experience today because, instead of repeating the stuff that I went over in the screencast, we kind of had a collaborative game design experience. I opened up the application and asked the students, "OK, we need to start with a room. What shall we call the room?" Somebody called out "International House of Puppies." That set the stage for a rather silly, surreal adventure... including features like a chest full of pancakes, I.F. cliches like locked doors, and flesh-eating zombies. It was weird, but the students really got into it, and had a good time. I had a hard time warding off all their questions.

Inform, I think, is going to be a lot more difficult. I've read all of the documentation (well, I skimmed those last few chapters, anyway), and it's a lot more... subtle... than Adrift is. It does use programming, but not what I'm used to. It uses natural language. Although I use the term natural very loosely.

I'll let you know how that goes. The workshop for Inform is on Wednesday...

Well, so much for the garlic theory.

On the day that Heather and I left for Chicago, I started getting sick again. So, naturally, I thought it would be a good idea to test if garlic worked in reducing symptoms. Throughout the whole weekend, I ended up choking down about eight cloves, in total (Heather has since virtually forbidden me to eat any more). The symptoms did get a little better, but it was nothing like what I experienced earlier - a complete disappearance of all symptoms. So, in the end, raw garlic is not a cure-all.

Actually I got pretty ill. And then Heather did. Neither of us is completely over it -- I'm still coughing a lot, and my nose runs occasionally -- but it's a lot better. I guess it's just one of those things, after you return home from being abroad for a while. Every time I returned I seemed to get really sick. Oh well... at least no pneumonia this time. Yet.

Chicago was good. It was great to see everyone. Also, it was good to see my Uncle Dennis. It had been a long time since I had last seen him. We ate at a good Mexican restaurant called Mi Tierra. Yummm! Anyway, Uncle D. looked great, and we had a really good conversation. Among other things, we talked about Japan, and how the samurai spirit still resonates among Japanese workers today; the Japanese novel The Tale of Genji - which I still have not read, lamentably; and philosopher David Hume. Heather remarked to me after we left, "your family never talks about light stuff at dinner, do they?"