RECOMMENDATIONS IN E-LEARNING: IDENTIFYING CURRENT TRENDS WITH AUTOMATED CRAWLING METHODS
IMA, RWTH Aachen University (GERMANY)
About this paper:
Conference name: 13th International Conference on Education and New Learning Technologies
Dates: 5-6 July, 2021
Location: Online Conference
Abstract:Due to research and technological advancements, it has become increasingly challenging to keep up with trends. This concerns innovation-driven fields like e-learning in particular. Emerging trends are reflected in the digital world through scientific publications and blogs, written or informed by experts in the field. Following this variety of sources becomes a large time investment for educators with an ambition for innovation. As technologies, however, offer the opportunity to crawl the web and identify currently discussed topics, the question arises of how such technological solutions might be used to automatically detect current e-learning trends, without requiring human input.
The project ELLI 2 aims at supporting university teachers in an increasingly digitalized teaching environment. One goal of the project is to create an automated process that regularly compiles a list of the top five current topics in the field of e-learning. The challenge is to:
(a) identify relevant topics of high quality that are
(b) new and trending.
To accomplish this task, we combined two approaches. The first one bases on a trend detection method developed by Twitter that is called chi² static to identify current trending topics among a sample of blog posts from two popular e-learning blogs. We combine this method with a Wikipedia crawler that starts with relevant articles on e-learning and models linked pages by means of a connected graph. Based on the identified topics of Wikipedia, the script thereby compiles a list of the recently most mentioned topics of e-learning. This final list contains terms popular enough to be listed on Wikipedia, which have recently been mentioned in blogs.
The paper describes the approach for automated trend detection in the e-learning domain as well as its results and provides insights into how to integrate this method into digital tools, e.g. an e-learning recommender. Furthermore, it discusses the current shortcomings of the approach together with possible solutions to increase the scope and relevance of the identified trends. Despite its limitations, the proposed method provides a promising application of an existing trend detection technology, contributing to automated trend detection in general and in the field of e-learning in particular.
Keywords: Trend detection, e-learning, Wikipedia crawler, digital teaching, recommender system.