Area 51 » Science

followers

Follow It!

20 Example Questions (1 closed)

active oldest votes

up vote 12 down vote

added by lazyCrab Nov 28 '12 at 19:17

Need to be specific in asking which algorithm for what task. – 2er0 Jan 29 at 17:50

up vote 11 down vote

added by Joel Cornett Nov 27 '12 at 12:13

e.g. pos, sense-tags, domain-labels. – 2er0 Apr 5 at 6:10

up vote 10 down vote

added by Joel Cornett Nov 27 '12 at 12:11

up vote 10 down vote

added by THelper Dec 17 '12 at 8:33

	`Frog` (ilk.uvt.nl/frog) is one of the simpler dependency parses for Dutch, that i know of – 2er0 Apr 5 at 6:09
	@2er0 We are looking for good example questions to determine the scope of new stackexchange site on NLP. Answering the questions is not the goal right now. That should be done when the site is launched. – THelper Apr 5 at 7:14
	@THelper, yeah, but it's still nice to get some answers around to know that the example questions are good, no? Also it's a pretty good software too, so `Sharing = Caring` =) – 2er0 Apr 5 at 7:27

up vote 10 down vote

added by Philipp Claßen Jan 4 at 14:19

up vote 10 down vote

added by bendaizer Jan 21 at 10:44

indexing of what? – 2er0 Jan 29 at 17:50

indexing of texts as vectors in a semantic space vector approach, for Latent semantic analysis, random projection technics etc – bendaizer Jan 30 at 9:36

up vote 10 down vote

added by Alexey Romanov Feb 4 at 12:07

	E.g. `Mr. president of U.S.S.R` shouldn't be segmented into `Mr`, `president of U`, `S`, `S`, `R`. – 2er0 Apr 5 at 6:11
	I'd expect any SBD algorithm to do well on this. I meant, e.g. detecting that `The president of Russia is Putin He is a human` should be segmented into `The president of Russia is Putin`, `He is a human`. – Alexey Romanov Apr 5 at 8:01

up vote 10 down vote

added by Wesley Baugh Feb 28 at 22:41

That would be a summary of Robert Navgili's survey paper =) – 2er0 Mar 6 at 0:30

up vote 7 down vote

added by Joel Cornett Nov 27 '12 at 12:13

closed as not a real question by Robert Cartaino♦ Apr 5 at 17:47

It's difficult to tell what is being asked here. This question is ambiguous, vague, incomplete, overly broad, or rhetorical and cannot be reasonably answered in its current form.

this isn't a question? – warren Nov 27 '12 at 14:54

up vote 7 down vote

added by Wesley Baugh Feb 28 at 22:18

up vote 6 down vote

added by 2er0 Apr 5 at 6:05

up vote 4 down vote

added by 2er0 Apr 5 at 6:03

up vote 4 down vote

added by Hamman Samuel Apr 18 at 13:34

up vote 3 down vote

added by Sai Manoj Kumar Yadlapati Mar 25 at 16:49

Firstly, how many tokens/words are then in each record? Next, what is the task you are using the n-gram for (the value of n is also dependent on the task)? Lastly, how much computing juice do you have (to handle a googlian-gram is gonna kill my laptop for sure)? – 2er0 Apr 5 at 7:31

Can't express all of them in the question due to limitation on no. of characters. – Sai Manoj Kumar Yadlapati Apr 6 at 12:17

@2er0: Hope it is okay now? – Sai Manoj Kumar Yadlapati Apr 7 at 10:10

up vote 2 down vote

added by 2er0 Apr 5 at 6:06

the process to mkcls then plain2snt then snt2cooc then make gizacfg file, then GIZA++ is a real hassle. I am sure someone must have scripted all that and just take in 2 sentence aligned file as inputs with several optional parameters for giza configuration – 2er0 Apr 5 at 6:08

up vote 1 down vote

added by 2er0 Apr 5 at 6:02

up vote 1 down vote

added by 2er0 Apr 5 at 6:04

It would depend on the wording of the actual question and how much research effort the asker can show. It treads the line between reference request (ok) and "show me how to do this" (bad) – acattle Apr 30 at 16:42