An Optimization Approach to Automatic Construction of Browsable Concept Index for Organizing Online Educational Content

As learning becomes more dependent on online learning platforms, new challenges emerge. One major challenge is efficient access to educational content as most content is unstructured and scattered on the Web. Indeed, there is a great need for organizing educational content and presenting it in a semantically structured and browsable format that does not require the users' prior knowledge of relevant content. In this paper, we address this need by introducing a novel semantic structure, called Browsable Concept Index (BCI), to bring semantic structures to educational content. A BCI consists of a set of representative concepts that form a graph with each (weighted) edge indicating strength of association between two concepts and each concept node being linked to the text segments in the collection of documents that cover the concept. The BCI effectively integrates scattered textual content and enables efficient access to the content by users in multiple ways including: semantic browsing, exploratory and targeted information seeking, and collaborative information seeking. We further propose an optimization-based algorithm to automatically create a BCI based on a collection of educational textual content. Experiments on Coursera course content demonstrate that the our algorithm is effective, allowing the automatic construction of a useful BCI that can facilitate online learning.