Adaptive scenarios and tutoring using fuzzy reasoning in an intelligent educational adventure game
Προσαρμοστικά σενάρια και διδασκαλία με χρήση ασαφούς συλλογιστικής σε ένα ευφυές εκπαιδευτικό παιχνίδι περιπέτειας
KeywordsFuzzy logic ; Fuzzy reasoning ; Adaptive educational games ; Adventure games ; Intelligent tutoring systems ; GDPR and smart educational games ; Abstracted learning mechanism
This thesis describes an innovative educational prosocial digital game that allows the dynamic adaptation of scenarios of games tailored to the learning needs of students who are also players of the games. This is achieved by the employment of a user modelling component that identifies the level of knowledge of students using fuzzy reasoning and as a result it performs the dynamic prolongment of playing motivation by extending the game scenarios while the game is being executed. The dynamic adaptations of scenarios involve revisions and more detailed tutoring content as well as personalised exercises in the context of new plot situations of adventure for the game players. Another important novelty of the research described, is the abstraction of the student modelling component to be applied in a diversity of games. This abstracted mechanism allows the educational adaptive learn-and-play environment to be applied in other types of game, such as a two-dimensional adventure game, where the plot involves quests and a virtual reality game, where the plot involves playground battles. Educational digital games offer students significant benefits due to their immersive features that increase motivation for learn-and-play activities. This is particularly the case when educational games can adapt their content to the user's needs automatically. However, most educational games have very limited adaptive capabilities. Their adaptive features only apply to the educational aspect and do not consider the pleasure factor for which they were initially designed. It is crucial to give due importance to the entertainment factor alongside learning; otherwise, the whole process of educational entertainment will be unsuccessful. If an educational game neglects the enjoyment part and does not provide mechanisms to alter its engaging elements individually and dynamically, there is no point in using it in the learning process as the student will quickly get bored and abandon the game. Hence, an educational game must recognise the student's educational needs and dynamically adapt its content at both the instructional and game level. In this way, it will provide individualised learning and gaming experience and keep the player's interest high enough to continue learning and playing as long as the game lasts. Another significant issue that researchers have recognised is that each player has their preferences regarding the type of game they enjoy. The student can play various games to practice their knowledge of a specific subject. However, each game contains different educational material and different or non-existent techniques for adapting the educational content to the student's needs. Therefore, a significant gap emerges in the field of educational games, according to which a generalised, abstracted, and available for each game mechanism for personalising education is absent. Incorporating the same personalised educational content in different games makes it easier for the student to choose the game they like, as the only criterion for the selection is the unique features of each game that appeal to the student. Selecting the player's favourite educational game leads to an engaging learning and playing experience and better educational outcomes. Given the above, this thesis describes an innovative approach that uses a fuzzy logic technique and combines two rule-based systems for the dynamic personalisation of the learning content and the game plot. The first rule-based system approaches the user from the student's perspective and determines the difficulty level of the tutoring content and the respective exercises. The second system treats the user from the player's perspective and decides to extend the scenario of a game by adding new game elements and tasks. Both systems base their decisions on the user's knowledge level. Thus, the student's knowledge is the basis for adapting the student's educational objectives and the game scenario. This approach, therefore, responds to the challenge in educational digital games regarding how a game can offer an individualised, engaging environment to keep players highly motivated based on their educational status and enhance the learning benefits. The thesis additionally describes a mechanism for gaming and learning where the abovementioned AI-based personalisation process has been generalised and abstracted to be available for use in games of different platforms and categories. It exploits web services, a widespread technology used in modern software development that supports data exchange between applications, to make the student model and the personalised content accessible in any game. This innovative mechanism is critical as any game can utilise it, and subsequently, the player can choose the game they prefer for their education, accessing the same personalised educational content. The gain of this novel approach is that each student enjoys an exciting gaming and entertainment experience and acquires the desired knowledge by playing their preferred game. The current research consists of two fully designed, implemented, and evaluated educational games of different types, a 2d adventure game and a virtual reality exploration game, for teaching HTML language. The research findings demonstrate two facts. First, when a game contains high levels of adaptivity by modifying its educational material and scenario according to the student's educational needs, it increases the player's enjoyment and significantly improves their learning performance. Secondly, when students have access to different games with the same individualised educational content, they can use the game they most enjoy and maximise their learning experience, mastering the desired knowledge more effectively.