This repository exclusively contains the Generation classes used by the VSCode extension. The alex branch of my fork of the ludii repository contains most of my work.
approaches.symbolic.SymbolMapper Maps symbols to the symbols which can be used to initialize them. It pre-computes all combinations of clauses with different optional parameters, and-groups and or-groups. Given a set of previously selected parameters, the possibilities for the next parameter are filtered on-request.
approaches.symbolic.nodes.* These 8 classes are used to represent the tree-like structure of ludii games. Each node corresponds to a symbol and is responsible for interfacing with the SymbolMapper to obtain possible child/parameter nodes. They are also responsible for compilation, which allows them to cache previously compiled ludeme branches when a ludeme tree is modified. They make use of the SymbolMapper
approaches.symbolic.PartialCompiler Compiles possibly broken game descriptions into a tree of GenerationNodes. It's advantages over the standard compiler are that:
- It returns partial compilations states. Meaning that a game generator can recover by replacing a problematice ludeme without re-compiling the whole game.
- It reppresents games as trees of GenerationNodes, which enables quick alterations to the ludeme tree.
- It's slighly faster than the standard compiler, and there are still some valuable optimizations that could be made.
approaches.symbolic.api.* These 4 classes interface with the VSCode extension.
The main methods in the updated evaluation system can be found in EvalGames.java, which can be located in supplementary/experiments/eval/EvalGames.java. I have also added a new metric named Systematicity, which can be found in metrics/designer/Systematicity.java. The main methods that I have added are as follow (they are all commented, for further insight examine said comments):
-
getEvaluationScores - Simply evaluates a game. Provides multiple input and output options, so simplifies internal use of the evaluation function.
-
defaultEvaluationFast - Quickly evaluates a game with some tested settings.
-
defaultEvaluationSlow - Evaluates a game with some tested settings more thoroughly.
-
loadDB - Loads two pre-built CSV files into memory for later use.
-
recommendScore - Recommends an evaluation score for a game based its' nearest neighbors - either BoardGameGeek scores or metric scores can be used to do this. For this method, loadDB is necessary.
-
Firstly, since I'm using files that are larger than 100 MB I used git LFS to push it into main, so install git LFS and pull the large files using "git lfs pull" (Sorry for this)
-
To run my code run the Main in 'GUI/WelcomeDialogue' which is meant to introduce the user to the application
-
For Alex to integrate the code, you are able to access the recommended games in "GUI/ReccomenderStarter" in line 312, I presented it as a String titled desc_format which should be the descriptions in the format you want, I printed the output so you can see, let me know if I should change it.