Website founded by
Milan Velimirović
in 2006
0:16 UTC
| |
MatPlus.Net  Forum General what is the best online site for problem collections? |
| |
|
|
|
|
You can only view this page!
| | | (1) Posted by Eugene Rosner [Monday, Sep 21, 2015 15:42] | what is the best online site for problem collections?
Is it Schwalbe's PDB, yacpdb or perhaps another I'm not acquainted with? Tough to shell out the cash for winchloe, are the others up to the task?
| | | (2) Posted by Sarah Hornecker [Monday, Sep 21, 2015 17:08] |
I think the Schwalbe website could use a better interface, but apart from that those both are also the ones I know. There once was a project by Mihail Croitor, but it was only about studies. I am not sure if it still exists. Maybe someone should make a kind of YouProblem, where you can upload chess problems. :-)
| | | (3) Posted by seetharaman kalyan [Monday, Sep 21, 2015 19:22]; edited by seetharaman kalyan [15-09-21] |
Yacdp... has a superb interface and easy to use. Actually the same interface can be used to search in PDB also. Great work of Dmitri Turevski !!
| | | (4) Posted by Frank Richter [Monday, Sep 21, 2015 20:05]; edited by Frank Richter [15-09-21] |
It depends on your needs, and in my opinion the some Euros for WinChloe light are a very useful investment.
Only there you can really perform effective searches by keywords. In PDB is indexing on a very low level.
| | | (5) Posted by Eric Huber [Monday, Sep 21, 2015 20:22] |
@Siegfried:
Mihail Croitor's project is still alive, but the link has changed several times and it is difficult to keep track of it: http://elearning.usm.md/endgame/
| | | (6) Posted by Andrew Buchanan [Tuesday, Sep 22, 2015 07:28]; edited by Andrew Buchanan [15-09-22] |
It's healthy for numerous reasons to have multiple problem collections.
For me the most important factor in quality, is how comprehensive is the collection. One of the greatest sadnesses that can befall a composer is to find himself anticipated, despite all due diligence in searching for prior art.
In order to ensure that the hole doesn't get any deeper, owners of all the major databases should ensure that they ingest efficiently all problems from all relevant magazines. Of the editors, Christian Poisson on WinChloe understands how important this is, and has been energetic about reaching out to the magazines. I don't know YACPDB approach. I can say that PDB team do not seem to get this point, even for their own "in-house" magazine, Die Schwalbe. In this age of big data, RSS, etc etc it's crazy for a database to live in such a 19th century way. Since the data transfer is many-to-many, ideally there would be a standard format that magazines write to, and databases read from. It would be rational to have an open format based on the existing WinChloe transfer format.
It's about more than digging the hole deeper. The most likely place to find anticipation is in the most recently published material. So this really is a good 80/20 approach to fixing a quality problem and saving everyone a lot of time and effort.
The right first step would be for the owners of the databases to speak with one another, and see how much common ground there is, and whether a common solution is workable.
It would be good to if magazine editors are able to discuss and voice an opinion. If all problems published in magazines are hoovered efficiently into databases, where is the incentive for people to subscribe to the magazines and to read the articles? Maybe there needs to be a specified delay between publication and storage in a database? But still it seems very desirable to have a mechanism to ensure that all published problems do within a specified time period find their way into all major databases.
Thanks,
Andrew
| | | (7) Posted by shankar ram [Tuesday, Sep 22, 2015 16:28] |
"...existing WinChloe transfer format."
What is that?
| | | (8) Posted by Andrew Buchanan [Tuesday, Sep 22, 2015 19:52] |
> what is WinChloe transfer format.
I understand that a number of magazines make all their problems available to WinChloe. I hypothesize that it is consistent with Christian Poisson's overall approach that the files should be transferred in a single format to him. I have no directed evidence, but maybe I could start a chess magazine and wait for him to approach me :D
| | | (9) Posted by Frank Richter [Tuesday, Sep 22, 2015 20:54] |
I think the input/import to WinChloe is done manually by enthusiasts. You may refer the corresponding table on http://winchloe.free.fr/ouvrages_table_echecs.html ("Works" recorded in the table "Echecs"), where is written by whom (is this correct English?) the problems were recorded.
Some years ago Thomas Maeder tried to invite a standard problem format based on XML. But I think, there is no practical use yet.
| | | (10) Posted by Dmitri Turevski [Wednesday, Sep 23, 2015 07:01] |
I don't think the common format is the main difficulty here.
The problem is that magazine editors are pretty busy people and preparing the feed with original problems for each issue is extra work with no short-term and no mid-term return for them (and the long-term return is also questionable).
This thought is not mine, the topic was brought up during the computer matters subcommitee meeting in Bern.
I personally did contact one magazine editor regarding this issue some six years ago. The magazine - Mat Plus - already had have all the originals in standard format, all we had to do was to set up the feed. Milan was kind of enthusiastic about this idea at first, but somehow the process stalled (perhaps he had already decided to stop publishing originals and focus on MPR by then?)
| | | (11) Posted by Harry Fougiaxis [Wednesday, Sep 23, 2015 12:29] |
Andrew, the editors of magazines and organisers of formal tournaments do not send any data to Christian. As Frank pointed out, Christian and other enthusiasts input everything manually. There are exceptions here and there, for instance I asked Hans Gruber and Piotr Gorski to send to me the LaTex files of the 8th WCCT and of the Ostroda congress bulletin. Christian has implemented a utility which extracts only the positions out of the LaTex files and imports them directly to a WinChloe table; of course, the rest of the data has to be entered manually.
Dmitry, I can understand the unwillingness of magazine editors, but in my opinion they could easily prepare for each issue a text file with the FEN positions and make it available to the public. It would save us a lot of time and effort. We can get the rest of the data (composer's name, publication data, distinction, etc.) from the paper or PDF magazine itself.
| | | (12) Posted by Roland Ott [Wednesday, Sep 23, 2015 13:08] |
Just to make the list of chess problem data bases more complete.
There is also the Meson Chess Problem Database of Brian Stephenson on his BDS Website:
http://www.bstephen.me.uk/index.php/meson
| | | (13) Posted by Hauke Reddmann [Thursday, Sep 24, 2015 13:52] |
Input is hell.
For twomovers though, I can already tell you the following without being
forced to shoot you :-)
The Albrecht collection has been greatly expanded by the work
of Udo Degener and Wieland Bruch (et al.). It is widely thematically
searchable (!!). And I probably live to see the day when it goes
online officially.
Hauke
| | | (14) Posted by Eugene Rosner [Thursday, Sep 24, 2015 20:10] |
does the new APwin program have a search mechanism for its database?
| | | (15) Posted by Andrew Buchanan [Friday, Sep 25, 2015 06:30] |
A few random points, following up on the thoughtful responses in this thread.
- PDB has about 42,000 problems from "Sammlung Brian Stephenson". Is Brian's collection continuing to grow independently? Was the input to PDB a one-off exercise?
- I agree that "input is hell". There is an app https://play.google.com/store/apps/details?id=com.kgroth.chessocr&hl=en which allows for diagrams to be taken into an android device using optical character recognition. I haven't tried it out, because I have an iPhone myself. the programme links with game databases. What would be nice is for it to integrate with our problem databases, and there to be an easy way to input the basic metadata (author/source/date/stipulation). That at least would reduce to a minimum issues with data entry.
- Even without such an app, computer verification should be set to happen for most problems at the point of data entry, to flag that there is maybe a data entry issue, because at that point it's easy to check whether the diagram has been entered directly. So when I add a new problem to PDB, it will by default tell me whether it's C+ immediately.
| | | (16) Posted by Harry Fougiaxis [Saturday, Sep 26, 2015 09:24]; edited by Harry Fougiaxis [15-09-28] |
QUOTE
- PDB has about 42,000 problems from "Sammlung Brian Stephenson". Is Brian's collection continuing to grow independently? Was the input to PDB a one-off exercise?
At the moment, the Meson database includes 153,355 records, hence Brian has continued to work on the database. You should ask Brian and the PDB team if the input to PDB was an one-off exercise. Sadly, too many of those 42,000 problems have been imported without solutions and have not meanwhile been updated by the PDB team. Re. Meson, Brian acknowledges on his site that "the solutions have been generated automatically using a set of default options and are only present for #2s and #3s. No other stipulations have been tested and neither have any twin positions. Solutions may include tries that aren't thematic and lack tries that are." In general, the format of the solutions in Meson is at least very odd.
QUOTE
- I agree that "input is hell". There is an app https://play.google.com/store/apps/details?id=com.kgroth.chessocr&hl=en which allows for diagrams to be taken into an android device using optical character recognition. I haven't tried it out, because I have an iPhone myself. the programme links with game databases. What would be nice is for it to integrate with our problem databases, and there to be an easy way to input the basic metadata (author/source/date/stipulation). That at least would reduce to a minimum issues with data entry.
I tried ChessOCR in the past. You can scan diagrams from paper with your Android device and you get the FEN position, which can then be sent to a chess engine for analysis (or stored to a PGN file). You cannot input any other metadata. It works, but you should be careful how you scan otherwise it keeps on trying and trying... In the end, you spend more time in scanning than inputting the position manually by yourself.
There are two other OCR applications for Windows. DiagTransfer, http://alain.blaisot.free.fr/DiagTransfer/English/home.htm, can scan diagrams from an electronic file (PDF for example) or a web page and then you guide the program about the nature of the pieces. In the end you get the FEN position. Unfortunately you have to repeat the exercise for each diagram, so in the end you spend too much time; the program is in fact not so useful.
Much better is ChessBoardCapure, https://sites.google.com/site/fredm/#downloads The tool can learn a chess font after you capture 3-4 diagrams, so it is a real OCR application. As the programmer says in the tutorial https://www.youtube.com/watch?v=VUKfJkeCJmI, sadly the tool fails when you scan diagrams with 'hatched' black squares. Nevertheless it is a great application when it works and I have been using it a lot with very good results.
QUOTE
- Even without such an app, computer verification should be set to happen for most problems at the point of data entry, to flag that there is maybe a data entry issue, because at that point it's easy to check whether the diagram has been entered directly. So when I add a new problem to PDB, it will by default tell me whether it's C+ immediately.
I am not familiar with the data entry in PDB, but it seems that solutions are entered manually(?!) I have this impression because I have lately come across with entries marked as C+ even if they are ruined by major duals in main variations.
| | | (17) Posted by Andrew Buchanan [Wednesday, Sep 30, 2015 11:33] |
Thanks for your suggestions, Harry.
Chess Capture looks interesting, but I couldn't access the Chess Capture software without Norton security protesting about the page. I have emailed the author.
As far as PDB chess engine is concerned. I think back in the day, there was an engine run against the whole database, and that assigned values of C+ or cooked. I don't know how accurate it was. But these days, when a problem is entered, there is no automatic engine run against it. So folk may run an engine manually against the problem, and report their results.
| |
No more posts |
MatPlus.Net Forum General what is the best online site for problem collections? |
|
|
|
ISC 2024
