Intelligent Tutoring System for Sudoku with learning analytic capability

Abstract

Sudoku has been a very popular logic-building game for decades, yet there are limited resources for researches and teaching. Even there is already Intelligent Tutoring System (ITS) on different platforms such as online or mobile application; they are not as effective and dynamically helpful enough for players with different background to learn it. With the popularity of big data and artificial intelligence, data of the ITS could be kept and be further analyzed. Any trend or hidden correlation corresponding to the users' abilities and their performance may be useful for providing a more accurate learning environment. In this project, I proposed an ITS for Sudoku on Android platform that would perform data analysis. The primary feature of application is to capture data such as their background and learning behaviour from users and be able to analyse them to provide a more personal and customized game environment to users. Moreover, a better user interface with be created with more language options and simpler hints explanation. The system proposed consists of four major components: a Sudoku ITS that rests on Android platform, an external database that keeps track of all data collected from all registered players, a web server that connects between the application and the database, and the data will be extracted from the database, and fit back to existing users after being processed. Among all components, the Sudoku ITS is a vital part which interacts with and gets data directly from users. It is composed of an expert model, a student model, a tutor model and an user interface - a classic ITS model. While for the data analysis part, large amount of data collected will be analyzed through different machine learning methods. For example, Naïve Bayes Classifier, will classify the user's ability simply by data collected in the registration page and on the other hand , the Artificial Neural Network (ANN) will check if the hint is effective. It is a popular human-brain-imitating model which can perform different classification.