Arne Skjærholt

• About: Norwegian, computational linguistics student, classical philologer.

Recent Actions

• Commented on The syntax of English is undecidable
  As a full-time computational and sometime regular linguist, I feel the need to comment here. As several people have pointed out, the error here lies in the conflation of syntax and semantics. The syntax of (U) is perfectly valid. It...
• Commented on Data-Mining a Diagnosis in the ICU
  Coming from a somewhat related field (computational linguistics) I think in practice it'll turn out to be mostly a non-issue. Inference is a well studied subfield of statistics, and both computational linguistics and computational biologists mostly spend their time stealing...
• Posted The joys of backtracking to Arne Skjærholt

  I'm pleased to say that Parrotlog is now at a point where it can actually do stuff. It can call predicates and backtrack (more on that in a moment). Unification probably works as well.

  Now, backtracking is tricky to get right. Currently Parrotlog has a problem with cuts. A cut is supposed …

• Posted Decoding HMMs in Perl 6 to Arne Skjærholt

  I've wanted to write a reasonably useful Perl 6 module for a while, and I finally realised that the Viterbi algorithm would be a pretty simple place to start (hopefully it'll be useful as well).

  There's a module with the same name on ="http://search.cpan.org/~koen/Algorithm-Viterbi-…

• Commented on Dates in Latin
  I think Caesar would disagree. The old-school Latin designation for the 20th of June would be (I think) "a.d. XII Kal. Jul.", literally "12 days before the Kalends of July". I never did get along with that particular scheme... And,...
• Posted Parrotlog - Parsing Prolog to Arne Skjærholt

  For the syntax and semantics of Prolog, Parrotlog is based on a draft of the ISO/IEC Prolog standard (seeing how the actual standard costs muchos dineros).

  Now, the good news are that the Pr…

• Posted Parrotlog - Backtracking to Arne Skjærholt

  Backtracking is probably the defining feature of Prolog. Abstractly, the execution of a Prolog program can be seen as traversing a tree, looking for a path that fits certain criteria. Each node represents a choice between several options, and are usually referred to as choice points (for reasons…

• Commented on Parrotlog - Unification (again)
  Yeah, that would make a lot of sense. Thanks!...
• Posted Parrotlog - Unification (again) to Arne Skjærholt

  In the I went with the simple solution to the problem of unification. Variables point to other variables, and in the end a variable either points to a term, or nothing. What happens when you unify X with Y, Y with Z, and Z with X should probably not be considered just yet, and will probably have…

• Commented on Parrotlog - Unification
  Or maybe chaining the variables together and following the chain is the only way to do it? Hmm. Time to sleep on it, methinks......
• Posted Parrotlog - Unification to Arne Skjærholt

  I've finally hit the first real roadblock in the development: unification. Instantiating a free variable and unifying that with a term is simple enough (and works). The problem is when you start unifying free variables with each other. For example you have three free variables: X, Y, Z. You unify…

• Commented on Parrotlog - Getting started
  Yeah, lists is something that I've been pondering a bit. For a first stab I think I'll just represent all lists with ./2 internally, but a more efficient representation might make sense in the more long-term perspective....
• Posted Parrotlog - Getting started to Arne Skjærholt

  The inspiration for this project is primarily due to my reading On Lisp by Paul Graham, which talks about implementing non-deterministic search (or backtracking, if you will) with continuations (chapter 2022). Since I…

• Commented on Parrotlog - Prolog on Parrot
  The technical reason for Parrot is continuations. I stumbled upon a book called On Lisp (by Paul Graham), which implements a Prolog using continuations to do the backtracking. Another reason is that I'm already a bit familiar with Parrot from...
• Posted Parrotlog - Prolog on Parrot to Arne Skjærholt

  I've always liked Prolog as a programming language. Something about the way the language works (like the built-in backtracking) appeals to my nature, much like Perl's proclivity for linguistic floridity. Thus, I've decided to try my hand at writing a Prolog compiler for the Parrot VM.

  To t…

Comment Threads

• David Cantrell commented on Dates in Latin

  If you're going to use Latin, then the year is MMDCCLXIII (AUC).

• Burak Gürsoy commented on Dates in Latin

  Good catch :)

  Congrats Ovid.

• Jeffrey Kegler commented on The syntax of English is undecidable

  @matthias, Nathan, Arne: That syntax cannot be determined without semantics is something I first learned from Eugene Charniak, now at Brown, then at Yale.

  You are correct in finding this point "crucial". My demonstration reli…

• Andrew Rodland commented on The syntax of English is undecidable

  Meaning informs parsing. Meaning, and prior experience, clearly have a big impact on how our brains interpret a series of words. But it's not absolutely necessary. We can see unfamiliar words, intuit a part of speech for them, and move on. We can take deliberately constructed nonsense and come up with a parse for it. We can read and enjoy Jabberwocky. The parse may be more ambiguous when you have no idea what the utterance actually means, but it still seems clear to me that underlying structure provides part of the information, and semantics another part, and that the…