- The following discussion is an archived debate of the proposed deletion of the article below. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.
The result was keep. J04n(talk page) 18:40, 10 March 2013 (UTC)[reply]
- Graph-tool (edit | talk | history | protect | delete | links | watch | logs | views) – (View log · Stats)
- (Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL)
Software with no third party reliably published sources that cover it in nontrivial detail and would allow it to pass WP:GNG. There is also no other evidence of notability, and past tags requesting evidence have been removed without improvement. I removed one footnote from the article before taking it to AfD [1] but it does not mention the subject at all and does not even adequately source what it was being used for (the article's claim that a particular programming style provides large speedups). —David Eppstein (talk) 05:40, 21 February 2013 (UTC)[reply]
I have reinserted the link to the scipy performance comparison site. This site is authoritative, it addresses the questiion _directly_ that numerical code implemented in pure python can be orders of magnitude slower than pure C++ (just read the page carefully, including the summary table at the bottom). Furthermore this is _utter_ common sense, and is the reason why projects such as Numpy exists. Here is an excerpt from the Numpy wikipedia article:
Because Python is currently implemented as an interpreter, mathematical algorithms written in it often run slower than compiled equivalents. Numpy seeks to address this problem for numerical algorithms by providing multidimensional arrays and functions and operators that operate efficiently on arrays. Thus any algorithm that can be expressed primarily as operations on arrays and matrices can run almost as quickly as the equivalent C code.[1]
The only citation there is the same one which you had deleted. Please be consistent.
Would you care to elaborate why such similar software as NetworkX and Gephi is considered 'notable', but graph-tool isn't? executive_override (talk) 07:44, 21 February 2013 (UTC)[reply]
- Re NetworkX and Gephi, Please see WP:WAX. Re the disputed source, the issue is not whether general speedups to Python are possible: the issue is whether those speedups apply to this specific piece of software. Citing a source that talks about speedups elsewhere, but then implying without sourcing that the same speedups apply to this code, is intellectually dishonest and a violation of WP:SYN. —David Eppstein (talk) 07:55, 21 February 2013 (UTC)[reply]
- Just a note on this: The article has now been changed to reflect the source in a more unambiguous way: "This type of approach[1] can confer a level of performance which is comparable (both in memory usage and computation time) to that of a pure C++ library, which can be several orders of magnitude better than pure Python." It should be clearer now tha one is referring to a general approach. executive_override (talk) 09:41, 22 February 2013 (UTC)[reply]
You could be more explicit on your position on graph-tool vs NetworkX and Gephi, and so on. It would be relevant for the discussion.
Regarding the speedups, the source does not address it in a specific case, but in a general one. It is therefore a relevant source, and the claim made in the article is very reasonable. If you require a specific _proof_ for graph-tool, the claim could be reworded instead of the source removed. I cannot provide evidence for the claim myself, since it would not be third-party.
Here is a couple of articles on graph-tool: http://nethedz.org/blog/2011/07/installing-graphtool-on-snowleopard, http://jugad2.blogspot.de/2013/01/graph-tool-python-module-for-graph.html, http://tech-foo.blogspot.de/2013/01/visualising-ubuntu-package-repository.html. Also, it is included in the macports repository http://trac.macports.org/browser/trunk/dports/python/py-graph-tool/Portfile and freshports. executive_override (talk) 08:10, 21 February 2013 (UTC)[reply]
- Well, I didn't actually have a position or opinion on NetworkX or Gephi until just now. Our article here on Gephi is written in a way that makes it look notable; our article on NetworkX is not. But I just checked Google scholar and both of them have quite high numbers of citations to their documentation (188 for NetworkX, 226 for Gephi). That's, potentially, as many as 188 or 226 independent and reliably-published sources on these systems, although I would guess that many of them don't cover them in much detail. I don't see anything at all there for Graph-tool. —David Eppstein (talk) 08:47, 21 February 2013 (UTC)[reply]
- NetworkX is more popular than graph-tool, since it is much easier to install, being pure python (However, it is inadequate for high-performance computations, a gap which graph-tool fills). It also gathered more references, since they published papers describing the library, which make them citeable (and indexed by google scholar), whereas graph-tool is mostly informally used, and you don't have to cite if you use it. I've mentioned the references to graph-tool which I know of. If there is consensus that this does not qualify as 'notable', then go ahead and delete the page. I think it would be a pity, because it is an important element to the class of software which Wikipedia is aiming to document, since it includes NetworkX, Gephi and others. I would however like if the decision would be made by more than one single person. executive_override (talk) 12:02, 21 February 2013 (UTC)[reply]
- Note: This debate has been included in the list of Software-related deletion discussions. • Gene93k (talk) 14:21, 21 February 2013 (UTC)[reply]
Hi guys, I think you've been "misgoogling" this library. It happens often with terms generated from unspecific words connected by a hyphen, due to the way google interprets search input.
This search in google scholar returns 19 direct references to graph-tool in academic works. This is quite significant by comparison to the numbers provided above for gephi and networkx, given those two are easy-to-use packages intended for broader needs while graph-tool has both a more specific domain of excellence and is harder to put to use. (Also, though not really relevant, I searched by URI so all 19 are guaranteed to refer to this specific graph-tool, which I don't know is the case for the numbers mentioned for gephi or networkx.)
Sure, the article does beg for another citation in the text besides the one from scipy.org (which I judge pertinent as it states a universal fact about c++ code compared to pure python). The google scholar search I just provided might help with that. But I don't consider it a matter for deletion, not based on WP:GNG. I hope this helps settling this issue. =)
Cheers,
--Solstag (talk) 21:15, 21 February 2013 (UTC)[reply]
- Please note that I qualified that merge suggestion with
or other appropriate article
.
- I found that ref independently, rather than as a result of info posted elsewhere. It wasn't singled out.
- Cheers. -- Trevj (talk) 12:37, 22 February 2013 (UTC)[reply]
- I think a merge proposal must contain an appropriate target in order to be meaningful. Could you provide a mother article in which this article could be inserted?
- The article has now been updated to contain 11 academic texts (journal articles and academic theses), all of which qualify as reliable and independent. executive_override (talk) 13:02, 22 February 2013 (UTC)[reply]
- Keep. I'll state my opinion for the record. The article has now references to many independent and reliable academic sources which allow it to pass WP:GNG, and as far as I can tell no remaining issues raised in the comments above were left unaddressed. DISCLAIMER: I have a vested interest in this article. I have created it, and I am the author of the software. executive_override (talk) 15:04, 28 February 2013 (UTC)[reply]
- Relisted to generate a more thorough discussion so a clearer consensus may be reached.
- Please add new comments below this notice. Thanks, JayJayWhat did I do? 01:30, 1 March 2013 (UTC)[reply]
- The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.