A MODULE FOR INTELLIGENT ANALYSIS OF STUDENT MESSAGES TO SUPPORT PEDAGOGICAL ACTIVITIES DURING A DIGITAL LEARNING
Irina P. Burukina
Penza State University, Penza, Russian Federation
Keywords: digital learning, automation, text messages, architecture, neural networks, classification, optimization, workload
Abstract
The article focuses on the task of optimizing the workload of instructors involved in digital learning by automating processes related to exchanging emails with students. In particular, it presents a novel module for automatic analysis of text messages using machine learning methods. We propose a special deep neural network architecture adapted to the unique characteristics of student text messages. The system consists of a bidirectional recurrent neural network, an attention mechanism, and a dense output layer. We develop a module that successfully processes incoming emails from students, classifies them, and forwards them to the appropriate section of the forum located on the learning platform. The module also automatically generates answers to the most frequently asked questions. We demonstrate that the module helps significantly improve the pedagogical activity and reduces the time spent on responding to students’ emails. Using the module decreases the instructor’s workload and improves the students’ academic performance.
|
