Computer Systems Project Proposal
Justin Park
Period 1
Title: Development of a Better Algorithm For Go
Python
Oct 3/4 making simple game interface
Oct. 6 - on a field trip
Oct. 13 - found a protocol for GO AI.
Putting daily logs in folders on web page
Oct. 17 - gGo Game, needs to come up with a specific problem in this GO area.
Maybe become part of Gnu Go?
gnu go, gGo uses the AI from gnu go.
Oct. 24 - absent
Oct. 25 - absent
Oct. 27 - updated logs
Oct. 31 - backdated logs (last week) for sept - current
Nov. 1 - absent
Nov. 15, Tues - absent
Dec. 15 Current Status Update
I am looking at the source code of a Java GUI program (goGui,
goGUI) that reads and writes .sgf files (smart Game Format), which is the standard for saving
Go games. I am currently transcribing this into Python for my program.
Eventually this will be the method that I store games, and I will create a
database of these sgf files that I can easily access (it will be organized),
and refer to during the AI decision making process.
I am also looking at the official specifications for the protocol (at
sgf). Some of the interesting
features that may be helpful in AI coding is that it is a tree-based format,
meaning nodes of alternative moves can be stored. There is also the ability to
rank moves, a very important feature that can help differentiate good ones from
bad ones.
Dec. 20 -
I spent most of my time looking at bibliography references for the lab
requirements and continuing reading the protocol specs for the sgf
file format. For Mr Latimer: Here's a good tutorial for learning Go:
http://playgo.to/interactive.