A hard job indeed to remove all-but-one corrections, there are 2000  
names shown in the Spellcheck Results window and it doesn't go further  
than the letter C! My base contains more than 2 million games. Anyway,  
I checked some names, it seems to be a date-less problem. So I deleted  
all my games without date, and did the spell check again. It seems  
that I have to do it several times though.

So my next question, I used a one million base + enormous base + some  
of my GM collection and put them together to get after some cleaning a  
over 2 Million games. It seems that it is not very good way. I am  
mostly interested in ancient historical  games (ie Philidor), GM games  
and more. Does anyone would be happy to share his clean base or let me  
know where I can find a nice big one?

Charly


>> Thank you for your answer:
>> - I unchecked the Ambiguous box
>> - what I call the alert window is the window that pop up after  
>> making corrections and tells "xxx Player names occurring yyy times  
>> in total was corrected"
>
> Ok. Clear.
>
> Following will convince you of what I confirm below.
>
> In the script editor window, before pressing "Make Corrections,"  
> remove all-but-one corrections suggested. In this remaining  
> correction line there is a (<number>), indicating the amount of  
> games that scid is expecting to correct. Take a correction with some  
> serious number (not 1) and remember this <number>.
>
> Now do the corrections. My prediction is that in the alert box <xxx>  
> will be 1 (obviously!), and that <yyy> will be less than <number>,  
> possibly even zero.
> If this is the case you must have run into the date check, since  
> there is no other "possible" reason for this difference to occur.

>> - running the process again will find more names to correct!?
>> - for example the name "A Chernin" that should be corrected as  
>> "Chernin, Alexander" isn't and many more other names, it will  
>> appear again on the second search. Actually the concerned games are  
>> not dated. May be it is what you are explaining but I am still  
>> confused.
>
> Yep, this is the cause. And you can consider this another scid  
> deficiency with the date-ambiguity checking (but especially  
> signalling!), although I feel that your database is a bit crippled  
> as well. Chess games should have a date, especially if you want them  
> to assess developments in opening theory. I recommend you to refresh  
> your collection. :-)
>
> So far I did not even consider not-dated games. The logic (if there  
> is any...) behind the current implementation is that you can only  
> correct not-dated games played by not-dated players. :-)
>
> Try this: Open your ssp file. Remove Alexander's birth date (put it  
> back again later...), which is the '1960' after his [rating].
> Restart scid, and see that his games are corrected alright now...
>
> I am open to suggestions how to improve this or make it more flexible.
>
> In general, date checking reveals a real ambiguity very rarely.  
> Mostly  it occurs to games played by the not-so-well-known father of  
> the better-known son (named after his father), if the father is not  
> in the ssp file while the son is. The name search takes the two  
> persons as identical, the correction finds that son tried a few  
> games in his previous incarnation...
>
> On the other hand, I think scid should avoid making mistakes.
>
>>
>> I guess it is a tough problem to Spellcheck those player names as I  
>> could find more than a dozen of different spelling for a single  
>> player (ie Michael Adams, M Adams, M Adams <Eng>, M Adams <Eng, GM  
>> M Adams, Adams M, Adams, M, Adams, M., Adams, Michael and more  
>> combinations of these)
>
> Yes. It is a mess out there. You must have gotten games from all  
> over the place, since most game collectioners follow some standard  
> approach, that is recognized by scid.
> Of the once listed by you, most ARE covered by scid, fortunately,  
> although I am not sure of country names in <> pairs; in () pairs -  
> as used by convekta - is covered.
>
> Cheers,
> Joost.
>
>>
>>>>> Hi,
>>>>> I am using a Mac and report here some issues in a random order.
>>>>>
>>>>> 1) Spellcheck Player Names doesn't seem to work: when doing a  
>>>>> second
>>>>> search the same names to correct appear again, I guess that they
>>>>> weren't corrected despite that an alert window with the  
>>>>> corrections
>>>>> appeared.
>>>>>
>>>>
>>>> It works for me, are the names repeated preceded by the text   
>>>> "Ambiguous:" or all the names show up again?
>>>>
>>>
>>> Scid indeed tries to be careful with ambiguities. There are two of  
>>> those:
>>>
>>> (1) Ambiguous names are captured during the name search phase.
>>> These are player names in the Dbase that "almost-but-not-quite"  
>>> match alternative player names identified in your spelling/ 
>>> rating.ssp file.
>>> In a big Dbase there can be a lot of these and as it is in general  
>>> not the best idea to automatically correct ambiguous names, scid  
>>> has the option to hide these from the alert window (as you call  
>>> it) altogether.
>>>
>>> As an aside: The alert window is actually an editor, showing the  
>>> correction script that was generated by the name search. The  
>>> script syntax is described at the top of the text.
>>> If you want some ambiguous correction to happen anyway, you can do  
>>> so by modifying the associated script line (remove the text up to  
>>> " before the old name).
>>>
>>> (2) Ambiguous games are found during the correction phase.
>>> Suppose a name correction for player X is ongoing (original name  
>>> old_X) and the .ssp file identifies X's date of birth as May 2nd  
>>> 1974, then it may happen that scid finds a game in the Dbase  
>>> played by old_X in November 1967.
>>> Clearly, this game was not played by X (as he was not even borne  
>>> at that time) or the game date is wrong. Scid cannot tell, so scid  
>>> skips correcting the player name for this game.
>>>
>>> Once the correction is done, there still is at least one game  
>>> played by old_X. During the next name search, the name old_X will  
>>> pop up again...
>>>
>>>
>>> These are the more obvious reasons that can explain what you see.  
>>> If they do not explain, you may have run into a problem indeed.
>>>
>>> Cheers,
>>> Joost.
>>>
>>> <zip>


------------------------------------------------------------------------------
Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are
powering Web 2.0 with engaging, cross-platform capabilities. Quickly and
easily build your RIAs with Flex Builder, the Eclipse(TM)based development
software that enables intelligent coding and step-through debugging.
Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com
_______________________________________________
Scid-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scid-users

Reply via email to