Olivier Teytaud wrote:

> whereas I am looking for
> (2) input = a database D
>     output = frequently matched patterns

This is what I do:

I have thousands of separate SGF files and a program to merge
them all in a single binary file. This program filters: handicap
games, insufficient rank in players, fast time settings, duplicate
games, wrong board size, games with less than n moves, etc.

Some of these binary files I posted here previously and the one I
use most is in the supplementary materials of my IWCG 2010 paper.

Paper: http://www.dybot.com/papers/meval.htm
Supplementary materials: http://www.dybot.com/meval/

Then I scan this binary file very fast for:

* Full board patterns (fuseki)
* Corner/side patterns (joseki)
* Around the last move patterns (urgency)
* Statistics: e.g. How frequent is tenuki at move n, how frequently
  tenuki goes to the previous last move, how frequent is extension
  from atari at move n, or whatever I can figure out.

Now the bad news:

All the extraction is in thousands of undocumented Pascal (Delphi)
and x86 assembly source lines. It compiles for Windows. My mew engine
has new parts only in C++, but still a lot of assembly and Pascal. And
offline learning tools are mainly in Pascal.

It is not free software, but if you think it is interesting, I may be
very happy to cooperate.


Jacques.

_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go

Reply via email to