The widespread use of online learning object repositories has raised the need of studies that assess the quality of their contents, and their user¿s performance and engagement. The present research addresses two fundamental problems that are central to that need: the need to explore user interaction with these repositories and the detection of emergent communities of users.
The current dissertation approaches those directions through investigating and mining the Khan Academy repository as a free, open access, popular online learning repository addressing a wide content scope. It includes large numbers of different learning objects such as instructional videos, articles, and exercises. In addition to a large number of users.
Data was collected using the repository¿s public application programming interfaces combined with Web scraping techniques to gather data and user interactions. Different research activities were carried out to generate useful insights out of the gathered data. We conducted descriptive analysis to investigate the learning repository and its core features such as growth rate, popularity, and geographical distribution. A number of statistical and quantitative analysis were applied to examine the relation between the users¿ interactions and different metrics related to the use of learning objects in a step to assess the users¿ behaviour. We also used different Social Network Analysis (SNA) techniques on a network graph built from a large number of user interactions. The resulting network consisted of more than 3 million interactions distributed across more than 300,000 users. The type of those interactions is questions and answers posted on Khan Academy¿s instructional videos (more than 10,000 video). In order to analyse this graph and explore the social network structure, we studied two different community detection algorithms to identify the learning interactions communities emerged in Khan Academy then we compared between their effectiveness. After that, we applied different SNA measures including modularity, density, clustering coefficients and different centrality measures in order to assess the users¿ behaviour patterns and their presence.
Using descriptive analysis, we discovered many characteristics and features of the repository. We found that the number of learning objects in Khan Academy¿s repository grows linearly over time, more than 50% of the users do not complete the watched videos, and we found that the average duration for video lessons 5 to 10 minutes which aligns with the recommended duration in literature. By applying community detection techniques and social network analysis, we managed to identify learning communities in Khan Academy¿s network. The size distribution of those communities found to follow the power-law distribution which is the case of many real-world networks. Those learning communities are related to more than one domain which means the users are active and interacting across domains. Different centrality measures we applied to focus on the most influential players in those communities.
Despite the popularity of online learning repositories and their wide use, the structure of the emerged learning communities and their social networks remain largely unexplored. Our findings could be considered initial insights that may help researchers and educators in better understanding online learning repositories, the learning process inside those repositories, and learner behaviour.
© 2008-2024 Fundación Dialnet · Todos los derechos reservados