Enhancing English Language Assessment in Educational Settings using Natural Language Processing Techniques

Various approaches to testing the ENL learners in educational contexts have tended to experience some limitations concerning validity, reliability, practicality, and sustainability. These limitations can be well mitigated by NLP techniques especially in areas where automated assessment can be made a...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:2025 3rd International Conference on Intelligent Systems, Advanced Computing and Communication (ISACC) s. 438 - 443
Hlavní autori: Jayakumar, V Moses, Rajakumari, R., Alapati, Purnachandra Rao, Otero-Potosi, Santiago, Malleswari, D. Naga, Karthik, M.
Médium: Konferenčný príspevok..
Jazyk:English
Vydavateľské údaje: IEEE 27.02.2025
Predmet:
Accuracy
Automated assessment systems
Context modeling
Data models
educational settings
English language assessment
Feature extraction
Natural language processing
Natural Language Processing (NLP)
Personalized feedback
Sentiment analysis
Social networking (online)
Testing
Training
Tuning
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Various approaches to testing the ENL learners in educational contexts have tended to experience some limitations concerning validity, reliability, practicality, and sustainability. These limitations can be well mitigated by NLP techniques especially in areas where automated assessment can be made and personalized feedback given to learners. This particular work will contribute to the improvement of methods for teaching and testing in the sphere of English language by employing narrow components of NLP to create the base for the automated assessment that matches the need of a learner or academical curriculum and offers accurate and comprehensive feedback. In contrast with more conventional forms of assessment that involve paper grading or multiple choice tests the proposed approach leverages NLP to 'interpret' the student texts for the purposes of assessment thus encompassing multiple aspects of languages than can simply be graded out of out 100. The type of data used in the present study includes datasets like the Twitter Sentiment Analysis Dataset; this is a dataset that comprises labeled tweets that are used to train and test NLP models. An ideal approach, like BERT-CNN is used for feature extraction, preprocessing and training/validation to achieve higher levels of student outputs analysis. K-fold validation and parameter tuning strategies are used for effectiveness and overfitting avoidance of the proposed approaches. The proposed NLP-based assessment system will have a test accuracy of 0.99, and it will prove to be more accurate, efficient and scalable than traditional systems. The system shows its ability to fairly assess the language skills of the learners and offer the feedback to them; therefore, it underpins the potential for introducing changes in the language assessment in learning institutions. This is a major innovation in the area of English language assessment since this work provides a new approach of using NLP in education measurement. The proposed framework has potential for the betterment of learning assessment framework through automation of assessment and quality feedbacks all over the world for enhanced learning outcome of students.
DOI:10.1109/ISACC65211.2025.10969428