SQL Scrolls is an educational game developed as an extensible DGBL experiment to support knowledge acquisition in a database course through SQL training exercises. The game is used by students at the FHNW as a supplemental learning tool for the Database Technology module, however, due to a lack of reports regarding its usage patterns, it proved difficult to find ways in which to improve its effectiveness. The aim of this project was to facilitate improvements of SQL Scrolls by analyzing the current state of the tool to implement reports which can assist in understanding usage patterns.

Through the design and implementation of summary and usage reports, we provide an overview of the current state of the application that gives insights into the SQL concepts currently covered in the game exercises and information regarding game usage. Also, a Query Difficulty Taxonomy was developed to evaluate SQL queries within the SQL Scrolls application and assign a difficulty score. This was achieved by performing a literature review of existing research to identify known approaches for calculating the query difficulty.

A Query Difficulty Taxonomy was devised by adapting an existing published method which calculates difficulty based on the contents and attributes of an SQL query rather than analyzing usage data. To ensure compatibility with the content and nature of SQL Scrolls exercises, some adaptations to the existing method were necessary. To ensure the proposed solution was justified by scientific evidence, a further literature review was conducted to provide support in the definition of certain aspects of the approach (e.g. assignment of keyword weights). Following the implementation of the difficulty taxonomy as a script, the difficulty scores were compared to the average user attempts for each of the SQL queries to study the accuracy of the method. Additionally, two further approaches focusing on the lengths and contents of the query were implemented to serve as a comparison to the recommended approach. The comparison demonstrated that the proposed solution achieved the highest correlation between the difficulty score and the average user attempts for a query, suggesting that it is the most viable approach to determine the difficulty of SQL Scrolls queries.