wisskirchenj / ai-guessing

Simple AI-application, basic language processing, teaching a system to guess an animal

IDEA EDU Course

Implemented in the Java Core Track of hyperskill.org JetBrain Academy.

Project goal is to implement a simple AI-application (basic language processing) teaching the system to guess an animal. Tree structures, node traversals are used.

Technology / External Libraries

• Java 19
• Jackson Json, Xml and Yaml Serializing
• Internationalization (I18n) with PropertyResourceBundles in English, German and Esperanto
• PicoCli - a great CLI library
• Lombok
• Slf4j
• Tests with Junit-Jupiter and Mockito
• Gradle 8.0.2

Program description

The application will implement a simple interactive game where the computer will try to guess the animal, that the person has in mind with the help of yes or no questions. During the game, the computer will extend its knowledge base by learning new facts about animals and using this information in the next game.

Project completion

Project was completed on 26.03.23.

Repository Contents

Sources for all project tasks (6 stages) with tests and configurations.

Progress

27.02.23 Project started. Setup of build and repo with gradle on Kotlin basis.

06.03.23 Stage 1 completed. Simple language understanding, parsing and (random) reply generation.

06.03.23 Stage 2 completed. Ask for two different animals and a distinguishing fact, display learnings. First phrase creation and recognition.

11.03.23 Stage 3 completed. Full interactive guessing game, store learnings in binary decision tree, not yet persisted.

14.03.23 Stage 4 completed. Deserialize and Store the decision tree on game end and reload at startup. Allow storage formats JSON, YAML and XML specified by CLI-parameter and read in with picocli-library. Serialize Interface in tree node using @JsonSubTypes and @JsonTypeInfo-annotations

20.03.23 Stage 5 completed. Add a main menu - split GuessingGame and Controller. Add third controller class, that orchestrates the Knowledge Tree explore menu options. Implement 4 actions, that use - and inherit from - an abstract Depth First Search implementing class providing a process hook. Add TreePrintAction.

26.03.23 Final stage 6 completed. Full internationalization (I18n) support in English, Esperanto and German added, with basic grammar providing pattern support, implemented using PropertyResourceBoundle, java.text.MessageFormat and regexp-classes. Locale is set by System-Property via JVM-option user.language.