DispersiveWiki:Community Portal/Archive

Chopping up into smaller articles

I think we should split the big pages into smaller ones, to take more advantage of hyperlinking. At some point I'll also try to set up categories as is done in Wikipedia (which should presumably be our model for most things here. I've set up a rudimentary example of what I mean by this at the Other equations page, though many pages are still broken or not cleaned up at all. Terry 01:22, 27 July 2006 (EDT)

Categories

I've started putting in some categories into the Wiki; they provide a convenient way to dynamically index all pages associated to a single topic, and can themselves contain content of their own. Any page can have as many categories as it wants, so please feel free to design your own categories as you please. In the worst case they can always be merged or otherwise cleaned up. Terry 00:24, 29 July 2006 (EDT)

Redirects

Another useful Wiki feature is redirects. If a page contains nothing but the line

#REDIRECT [[some-other-page]]

then it will automatically redirect to some-other-page. This is useful for merging two otherwise redundant pages, or dealing with a concept that has many spellings or variant names, or with two topics which for some reason are being treated on a single page for now. Terry 00:24, 29 July 2006 (EDT)

Bibliography

What shall we do with the bibliography? It is hard to edit and hard to link to. I am thinking of moving the references to the page in which they are cited, instead of collecting them in one enormous file. This may cause a reference to be cited in more than one place, possibly with slightly different bibliographic data, but this should not be a problem. On the plus side, it makes it easier to find relevant references, and it should be relatively easy for a new user to update a reference or add a new one. Terry 01:22, 27 July 2006 (EDT)

Another possibility is to put each article on its own page. This could allow direct linking to the article and leave space for annotations and remarks on the article. One could also include links to mathscinet and arxiv. Colliand 16:55, 28 July 2006 (EDT)

This idea has advantages and disadvantages. One big advantage: because each article has a centralised location, any updates to the article (e.g. changes in publication status, or other annotations) only needs to be done once (this is important if the article is cited all over the wiki). Also, it will be easy for new users to update or otherwise edit an existing article. Disadvantage: it becomes harder for a new user to add a new article to the system. We may need to explain a step-by-step process (or have some sort of front end). Of course, new users can put down the references in any other way, and someone can always later move them to conform to whatever emerging standard we decide upon. Terry 00:24, 29 July 2006 (EDT)

There is also the issue of how to create a standardised naming system for these article pages. As you can see in the existing bibliography, I tried to assign two letters to each author and give a year, but this quickly led to all sorts of special exceptions and other headaches, and I didn't quite know what to do with preprints which did not yet have a year of publication. We could have an anarchic system in which each reference gets assigned whatever name the user who adds the reference sees fit to give, but this makes the references harder to locate, and also leads to the remote possibility that an article might be duplicated. On the other hand, it will be difficult to convince everybody to adhere to a complicated naming standard. Perhaps some sort of automation (either by an explicit front end for entering in references, or various cleanup robots) may help. Fortunately it is very easy to rename articles (and have links move accordingly) and so perhaps these issues will not be too serious, though it would be good to set down some minimal (evolving) standards to prevent utter chaos. Terry 00:24, 29 July 2006 (EDT)

Incidentally, I think the bibliography issue is the main issue we have to decide upon before the wiki is ready for wider distribution. Aside from minor corrections and for a few very enthusiastic editors (I hope we get some!), I expect the bulk of contributions from other PDE people to come from bibliography updates, so it would be good to have some sort of system in place for accomodating that first. Terry 00:33, 29 July 2006 (EDT)

OK, I think Jim's suggestion is probably the best one. For simplicity we can just use the existing reference names as the names for the wiki pages which contain the reference. Example: KnPoVe1989. I do like very much that we can now easily annotate each reference with commentary, links, etc. As an example of how the linking would work, see Kadomtsev-Petviashvili equation. I have also set up a category to hold all the reference pages, as well as some pages on naming conventions and reference management policies. What now needs to be done is the rather tedious task of moving each individual reference in the legacy bibliography page to its own page, as well as relinking all the citations. Perhaps some automated or semi-automated procedure would be better than doing this all by hand? Terry 15:38, 30 July 2006 (EDT)

Pieter Blue has fixed many of the links for the bibliography. I believe Pieter also knows an automated way to switch the bibliography items into separate pages. A different solution we should consider is to install the Biblio package. Colliand 22:16, 30 July 2006 (EDT) (Colliand 23:17, 30 July 2006 (EDT) Biblio collects data from the PubMed Identifier database. Can we use MathSciNet numbers to reference published articles and automatically generate the reference?)

I missed the discussion here and tried to simulate the old webpage as closely as I could. Each entry on the bibliography page is now a tier 5 subsection (=====) and can be linked to with [[Bibliography#key|key]] . Most keys are of the form {author abreviation}{year}{optional letter}. I'd suggest that we stick with this format, using capital letters, Au2005A, for preprints, and lower case, Au2006a, for published. A comment <!-- previously Au2005A --> can be used to record outdated keys. People can fix outdated links as they encounter them.

This combines the ability to link to bibliographic references, to annotate each entry, and to see all references on a single page. Of course, making each reference a page in the "Bibliography" category would achieve the same results. If we do decide to switch, it would probably be best to make the decision before chopping up the pages too much, since simple semi automated systems must be applied page by page. The advantage of the individual page system is that broken links are more obvious.

A few keys were not of the standard form, and a few links were previously broken, I'm going to try to fix all the links in the upcoming week. Pblue 23:05, 30 July 2006 (EDT)

It also looks like there were some pages which were not linked from the main page that I didn't change the format of, especially in the KdV section. Let me know if you find any. To update pages, in emacs, I replaced the regular expression "references.html#" by "references:" and "references:\([a-zA-Z]+[0-9]+[a-z]?\) \1" by "Bibliography#\1|\1". This replaces "references:key key" by "Bibliography#key|key"Nonstandard keys and broken links must be fixed by hand. Pblue 00:07, 31 July 2006 (EDT)

I like Pieter's suggestion of naming preprints by year (presumably of preprint release, rather than predicted publication date) and a capital letter. It is easy enough to move the page later once the paper is published, and update links as necessary. I also am now favouring Jim's idea of giving each paper its own page; it will be friendlier to update and to add new papers (e.g. one would plonk down [[Au2006a]] on the relevant page, click on the link, and add the reference, rather than writing [[Bibliography#Au2006a|Au2006a]], going to Bibliography, scrolling down until one finds the right alphabetical location, and entering in the new reference.) One does lose certain capabilities by fragmenting the reference files this way, but presumably the Bibliography category will be able to compensate for most of these. (One can also use ad hoc tricks such as searching for four-digit dates in order to locate references in a reasonably automatic manner.) In any case we should keep the old Bibliography file as a resource, at least until the porting is complete and the new system is stable. Terry 02:03, 31 July 2006 (EDT)

A quick wiki search for "references" reveals that 32 pages have old-style links. Oops, I guess we fragmented the old HTML pages a bit too quickly :) Still, it should not be too hard to clean it up. Terry 02:11, 31 July 2006 (EDT)

Linking to math reviews

Andrew Comech has written a tool called bibget. It does some neat things, e.g. UNIXPROMPT> bibget -r MR2121437 instantly retrieves the relevant math reviews reference in bibtex format. Colliand 00:07, 31 July 2006 (EDT)

Do we want to put references as [[Au2006a]] or [[Bibliography:Au2006a]]? Using the one page per reference section has the major advantage of allowing us to use the "what links here" link to hunt down all old links when a preprint is published, so I'm convinced this is the right way to go. Pblue 11:32, 31 July 2006 (EDT)

I think by fiat we are now using the former, as it is easier to type and looks nicer too. Presumably the Bibliography category will let us organise these pages easily even in the former case. Terry 23:59, 1 August 2006 (EDT)

I corresponded with AMS about linking to mathscinet from our bibliography pages. They directed me to the web page http://www.ams.org/msnhtml/getitem.html . So, for published papers in our bibliography we are free to link to mathscinet. For example, see http://tosio.math.toronto.edu/wiki/index.php/Bo1999b Colliand 12:31, 2 August 2006 (EDT)

Some reference pages, such as LiPo1993, have partial mathscinet links [1]. The string id at the end should be replaced by the associated math review number MR1249105. Is there an automated way to get these inserted into all the pages? Colliand 00:10, 3 August 2006 (EDT)

I am guessing that because we don't have specific tags for author, title, etc. one would need to build a reasonably smart system to automate things. Plus one expects references to continually be added by new authors, which we probably can't expect to adhere to any really complicated rules. Still a semi-automated system may be possible. Presumably once a mathscinet link is available one can query the AMS database in some automatic manner to aid our own somehow. Terry 00:39, 4 August 2006 (EDT)

Further correspondence with AMS from Drew Burton:

Getitem has the other nice feature that users without MathSciNet

licenses do get the cannonical bibliographic information and links to

the original article at the publisher's site when possible.

Another free tool that might be of interest is MRef for looking up

references in the MathSciNet database at http://www.ams.org/mref. It

takes normal reference strings and does an awfully good job of looking

them up in our database. It provides a variety of result formats.

E.g. I used it on your entry for BusPer1993 and got:

Buslaev, V. S.; Perel\cprime man, G. S. Scattering for the nonlinear

Schrödinger equation: states that are close to a soliton. (Russian)

Algebra i Analiz 4 (1992), no. 6, 63--102; translation in St.

Petersburg Math. J. 4 (1993), no. 6, 1111--1142 MR1199635 (94b:35256)

There is more information about our linking tools at

http://www.ams.org/tools in the "Linking" section.

It would be nice if we could write a link which takes the reference data on our bibliography pages already, uses that information to :search mathscinet, then displays the relevant mathscinet article listing. I don't know how to write such a link but will start to :experiment a bit. Colliand 11:19, 4 August 2006 (EDT)

In the long term, I think we should build a template for bibliography articles, with various arguments for bibliographic data such as author, title, etc. The advantage is that we can change the format of all articles simultaneously by editing the template. In the short term, the important thing is to get the basic data in the right place; data migration can then be done basically at any time. Terry 15:04, 10 August 2006 (EDT)

There are no, non-trivial links to the old Bibliography page. Pblue 10:52, 10 August 2006 (EDT)

There's a very basic template:bibliography. Pblue 13:41, 10 August 2006 (EDT)

Looks good! I tweaked it a little and set up one of my own references, Ta2004, using the template as an example (though I don't have access to MathSciNet right now). At some point we should migrate over but there is no urgency; it seems that one should search for a conversion process which is as automated as possible. Terry 15:04, 10 August 2006 (EDT)

Open for business?

The wiki is developing nicely. Not only is the conversion of the existing text reaching a point where the pages look nice (and invitingly easy to edit), the wiki format is also now suggesting some natural additions beyond the original brief of listing local and global wellposedness results for various equations. I particularly like the developing concepts category, making this wiki more encyclopediac (Dispersopedia, anyone?). Anyway, maybe we can start inviting some more people in on the project. Next week when I return to UCLA I think I will advertise it on my own web page. Terry 03:00, 31 July 2006 (EDT)

I have sent out a few invitations by email already. How about we arbitrarily declare Aug 7 "opening day"? :) Terry 00:39, 4 August 2006 (EDT)

Stubs

I have stolen the stub template from Wikipedia and adapted it here. A lot of smaller articles in need of expansion are now stubs to be expanded; see the stub category for a listing. It's easy enough to tag something as a stub, just put {{stub}} at the top of the offending page. Hopefully this will be one easy way for users to make significant contributions to this wiki :) Terry 01:07, 4 August 2006 (EDT)

Why not just use Wikipedia?

Why have a separate project? Why not just edit the standard Wikipedia?[2]Crust 14:56, 25 August 2006 (EDT)

I would imagine that this would probably upset many editors in the Wikipedia community. Looking at the policies on what Wikipedia is not, I would think that putting a large number of technical pages on Wikipedia (such as the 600+ pages of bibliography) which are really of interest only to specialists would be inappropriate. Perhaps in the far future there could be some sort of merging, though. Terry 11:32, 27 August 2006 (EDT)