?

Log in

Previous Entry | Next Entry

Title: The Next 700 Data Description Languages
Speaker: Kathleen Fisher (AT&T Research)
Date: Wednesday, March 8, 2006
Time: 10:00 AM
Location: Nichols 236
ABSTRACT

In the spirit of Landin, we present a calculus of dependent types to serve as the semantic foundation for a family of languages called data description languages. Such languages, which include PADS, DataScript, and PacketTypes, are designed to facilitate programming with ad hoc data, i.e., data not in well-behaved relational or XML formats. In the calculus, each type describes the physical layout and semantic properties of a data source. In the semantics, we interpret types simultaneously as the in-memory representation of the data described and as parsers for the data source. The parsing functions are robust, automatically detecting and recording errors in the data stream without halting parsing. We show the parsers are type-correct, returning data whose type matches the simple-type interpretation of the specification. We also prove the parsers are ``error-correct,'' accurately reporting the number of physical and semantic errors that occur in the returned data. We use the calculus to describe the features of various data description languages, and we discuss how we have used the calculus to improve PADS.

Joint work with Yitzhak Mandelbaum and David Walker.


Again, a nice talk, very informative, and naturally more in-depth than the general-audience talk. These specialized talks are usually a little sparser on attendance, but I think Fisher had at least 15 in this to two dozen or so in the other.

After lunch and coffee with the other faculty yesterday, I had a good talk with Fisher about hierarchical HMMs, IE, statistical learning of relational models, ontology extraction, and how hierarchical probabilistic models such as are used automatic speech recognition (ASR) relates to context-specific modeling of data. So, that was interesting. I'll post more on this later if anyone is interested.

We had another bioinformatics candidate, and of course I won't say anything here about his talk or about him until we are done with the interview process, but I do want to mention something in case he finds this blog and reads it. I had the duty of talking him to a local restaurant for dinner, and I had had a gradual headable building up from noon onward. It was mostly sleep/caffeine-deprivation induced, I thought, until I was feeling a dull throbbing by the time his seminar wrapped up. Well, we went to the Gold Fork across the street, and I ordered as small and fussless a meal as I could imagine: chicken penne alfredo with broccoli. Let me tell you, broccoli tastes like a whole other Brassica - in fact, more like pure cellulose, when you're coming down with what looks to be a 24-hour stomach flu.

I made the least contagious gestures I could, but I'm not as good as Donald Trump or David Heckerman about not shaking hands, even when I'm the carrier. Ugh.

So it was that I came home to shiver and suffer. I've been blessed with robust health and have been sick less than once a year throughout my life, probably about 3-4 days a year on average; there have been periods of more than 3 years when I have not been ill at all.

As I've said before, however, students have a distinctly anti-Pauling ethic: that is, whereas Dr. Linus Pauling advocated self-quarantine as a social responsibility, students tend to adopt one of the following stances:

  • 1. Immune systems need to be exercised, not protected. This may be the case, but there is a matter of degree when it comes to exercise, and IMO to say that people who are sick should just come right in and spread viruses and bacteria around at will is socially irresponsible. More important, we always have to ask, scientifically, what we are comparing: some exposure versus none, or massive exposure versus normal ambient exposure. The amusing George Carlin stand-up monologue notwithstanding, "that which does not kill me makes me stronger" is beside the point here. Unnecessary exposures at the height of contagiousness are superfluous at best, harmful at worst, and anyone who tries to justify selfish exposure of peers, customers, and students in the name of "doing them an immunological favor" is lacking in medical common sense and spewing bunk. So there.

  • 2. Professors need to see that I'm sick. No, thank you! A doctor's note will generally always suffice, and often enough, if you simply call in or send e-mail shortly in advance, it will do. If you're too ill to do so, a physician's note after the fact is the way to go. "Proof by inoculation" went out with the era of medical excuses for absenteeism being the only valid ones besides deaths in the immediate family. Nowadays, I have students cutting class just because they want Thursdays or Fridays off (don't think I don't notice!).
    My least favorite request for an extension is: "*kof* I need an extension, I'm *kof* sick *kof*". In that case, kindly submit your request through the Sublime Veil of IP! (Your classmates will thank you for it; not only can you keep from infecting them, an airborne TKO on the instructor won't help graded homeworks and exam guides get distributed faster!)

  • 3. I'll lose credit just by missing class. Listen: if you are a conscientious student who has come to class, done assignments, did not simply blow my course off because you were busy or apathetic, and really are feeling ill, then rest easy. My Tegrity recordings were made with you in mind. Watch the recording, get notes from your classmates or bring a USB drive to my class, and check with me to see what you missed. If you missed a rare in-class exercise, quiz or exam, I'll certainly let you make it up.


Moral: either way, just be responsible. Use the system, don't abuse it.


Summary: I'm shivering, I took my first acetominophen of the year and it's not even April yet, and the candidate either knows I'm sick or thinks something's wrong. If you're reading this: it wasn't you, it was the flu, and I hope you don't have it too.

--
Banazir

Comments

( 3 comments — Leave a comment )
mapjunkie
Mar. 12th, 2006 01:57 pm (UTC)
Yes, please: Hierarchical models and ontology extraction
I am interested in hierarchical learning and ontology extraction, so I'd be interested to read anything you have to post about that. Actually, I know RG is majorly interested in these topics as well, so if you ever get up around UC way, RG has a research talk series, and I'm sure Duane and I would be delighted if you were to give a talk on these subjects.
util
Mar. 12th, 2006 03:29 pm (UTC)
Where does Pauling talk about self quarantine? I'd like to hear more of what he has to say on this. Thanks.
util
Mar. 12th, 2006 03:35 pm (UTC)
Oh, and hope you feel better soon!
( 3 comments — Leave a comment )

Latest Month

December 2008
S M T W T F S
 123456
78910111213
14151617181920
21222324252627
28293031   

KSU Genetic and Evolutionary Computation (GEC) Lab

Teunciness

Breakfast

Science, Technology, Engineering, Math (STEM) Communities

Fresh Pages

Tags

Powered by LiveJournal.com
Designed by Naoto Kishi