| Fuad Mammadov, Huseyn Sultanli Development of IELTS Essay Evaluation System with Deep Learning |
|---|
| Abstract. The IELTS Writing Task 2 consists of well-written essays by non-native English speakers to be graded on four categories: Task Response, Coherence and Cohesion, Lexical Resource, and Grammatical Range and Accuracy. Manual grading of these essays is time-consuming and heterogeneous in nature, thus solutions through automation are required. This project gives an example of how a Long Short-Term Memory (LSTM) model, a sequence-trained recurrent neural network, could be employed to mark IELTS Writing Task 2 essays. The model, having been trained on a labeled corpora of essays, returns a total band score and in-depth feedback per criterion. Exploiting LSTM's ability to process text contextual dependence, the system is extremely human-like and accurate marking as possible. Metrics of performance such as prediction accuracy and processing time indicate its potential usability in real-time applications. It enables actionable, real-time feedback for student self-learning and aids teachers in low-resource settings. The project exemplifies automated essay marking, which proves the effectiveness of LSTM-based systems for edtech. |
| Keywords: LSTM, IELTS Writing Task 2, automated scoring, natural language processing, educational technology |
 Download PDF |
| DOI: |
