Ben Lambrechts wrote:
I don't want to hack in the code, but if I can create a new fuseki-
database,
I am willing to help.
I don't think a new full-board database would be of much value at this
time. As described in my other mail, smaller scale patterns have much
more potential.
However, I believe patterns/extract_fuseki.c, the code used to generate
the fullboard patterns, has a mode to also generate halfboard patterns.
It could be interesting to generate a collection of such patterns from a
set of strong games, which then could be used in an Elo pattern rating
approach.
Do you think it would be usefull to generate a new fuseki19.dbz from a much
larger game-database? I tried the first moves by hand:
F-H0-1 3977 pd
F-H0-2 1136 pc
F-H0-3 201 od
F-H0-4 158 jj
F-H0-5 137 qe
F-H0-6 101 qc
F-H0-7 46 oe
from the original file becomes
F-H0-1 31494 pd
F-H0-2 15635 qd
F-H0-3 782 oc
F-H0-4 633 nd
F-H0-5 609 jj
F-H0-6 482 od
F-H0-7 394 qc
F-H0-8 291 oe
with the database I have atm.
The fuseki19.dbz would become much larger, but it has much more strong
fuseki within it.
No, I don't think it would be very useful to generate a bigger
fuseki19.dbz, at least not useful enough to include it in the
distribution. But there are other possibilities. We could e.g. offer a
bigger fuseki database as a separate download.
It is also today relatively straightforward to test the effect of a
bigger fuseki database. Just put up two GNU Go which only differ in the
fuseki database on the 19x19 cgos server (see
http://cgos.boardspace.net/). If the effect is significant enough,
adding it to the distribution would be considered.
/Gunnar
_______________________________________________
gnugo-devel mailing list
gnugo-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/gnugo-devel
