A Review of the Trends and Challenges in Adopting Natural Language Processing Methods for Education Feedback Analysis

Thanveer Basha Shaik, Xiaohui Tao, Yan Li, Christopher Dann, Jacquie McDonald, Petrea Redmond, Linda Galligan

Abstract

Artificial Intelligence (AI) is a fast-growing area of study that stretching its presence to many business and research domains. Machine learning, deep learning, and natural language processing (NLP) are subsets of AI to tackle different areas of data processing and modelling. This review article presents an overview of AI impact on education outlining with current opportunities. In the education domain, student feedback data is crucial to uncover the merits and demerits of existing services provided to students. AI can assist in identifying the areas of improvement in educational infrastructure, learning management systems, teaching practices and study environment. NLP techniques play a vital role in analyzing student feedback in textual format. This research focuses on existing NLP methodologies and applications that could be adapted to educational domain applications like sentiment annotations, entity annotations, text summarization, and topic modelling. Trends and challenges in adopting NLP in education were reviewed and explored. Context-based challenges in NLP like sarcasm, domain-specific language, ambiguity, and aspect-based sentiment analysis are explained with existing methodologies to overcome them. Research community approaches to extract the semantic meaning of emoticons and special characters in feedback which conveys user opinion and challenges in adopting NLP in education are explored.

Introduction

In the modern educational landscape—ranging from traditional universities to Massive Open Online Courses (MOOCs)—student feedback is a primary driver for institutional improvement. However, qualitative feedback (textual comments) is notoriously difficult to scale. Manually reading and categorizing thousands of student opinions is time-consuming, subjective, and resource-intensive.

The challenge of "opinion mining" in education involves bridging the gap between raw student text and actionable insights regarding teaching practices, curriculum design, and learning environments. The core challenge lies in the complexity of human language: students use sarcasm, emoticons, and highly specific academic terminology that standard "off-the-shelf" sentiment analyzers often fail to capture.

Methodology and NLP Pipeline

The NLP pipeline for education involves several key methodological stages, moving from raw text to structured insights.

Feature Extraction and Selection

Traditional methods like Bag of Words (BoW) and TF-IDF focus on word frequency. Modern approaches utilize Word Embeddings (Word2Vec, GloVe) and Transformers (BERT), which provide context-aware representations.

To optimize model performance, feature selection is employed to reduce dimensionality. Two primary approaches are Filter and Wrapper methods:

  • Filter Methods: These involve ranking features based on statistical measures and selecting the top-performing ones before the modeling stage.
Filter methods rank and select features independently of the classifier.
  • Wrapper Methods: These use a predictive model to evaluate feature subsets, iteratively training and evaluating to find the optimal combination.
Wrapper methods use a classifier to iteratively evaluate feature subsets.

Topic Modelling and Summarization

Latent Dirichlet Allocation (LDA) remains the dominant unsupervised method for discovering hidden themes in student comments. For summarization, the field is shifting from Extractive Summarization (selecting key sentences) to Abstractive Summarization (generating new paraphrased text), driven by deep learning architectures like LSTM-CNN.

Key Insights and Results

The transition from "word-counting" to "context-understanding" is the most significant trend. Models like BERT are preferred for their ability to distinguish nuanced academic contexts. Furthermore, Aspect-Based Sentiment Analysis (ABSA) has become critical, allowing institutions to identify specific areas of concern (e.g., "lab equipment" vs. "lecture clarity").

Quantitative benchmarks from the literature show:

  • Accuracy: Deep learning models (CNN, LSTM) achieve sentiment classification accuracy between 86% and 94%.
  • Efficiency: BERT-based keyword extraction can reduce feedback text volume by 59% while maintaining predictive power.
  • Summarization: Hybrid models have demonstrated a 4.4% increase in ROUGE-1 scores over previous state-of-the-art systems.

Challenges and Future Directions

Despite progress, several bottlenecks remain:

  1. Sarcasm and Nuance: Sarcasm remains a frontier challenge, requiring multi-modal approaches to detect "sentiment flips."
  2. Data Imbalance: Educational datasets are often skewed toward extreme sentiments, biasing models.
  3. Interpretability: The "black box" nature of deep learning makes it difficult for educators to trust automated evaluations.
  4. Privacy: Balancing insight extraction with student anonymity is a persistent ethical concern.

Future research must focus on the pedagogical integration of these tools to ensure they lead to measurable improvements in educational outcomes.

Summary updated

Abstract

Artificial Intelligence (AI) is a fast-growing area of study that stretching its presence to many business and research domains. Machine learning, deep learning, and natural language processing (NLP) are subsets of AI to tackle different areas of data processing and modelling. This review article presents an overview of AI impact on education outlining with current opportunities. In the education domain, student feedback data is crucial to uncover the merits and demerits of existing services provided to students. AI can assist in identifying the areas of improvement in educational infrastructure, learning management systems, teaching practices and study environment. NLP techniques play a vital role in analyzing student feedback in textual format. This research focuses on existing NLP methodologies and applications that could be adapted to educational domain applications like sentiment annotations, entity annotations, text summarization, and topic modelling. Trends and challenges in adopting NLP in education were reviewed and explored. Contextbased challenges in NLP like sarcasm, domain-specific language, ambiguity, and aspect-based sentiment analysis are explained with existing methodologies to overcome them. Research community approaches to extract the semantic meaning of emoticons and special characters in feedback which conveys user opinion and challenges in adopting NLP in education are explored.
BibTeX
@inproceedings{shaik2022review,
  author = {Thanveer Basha Shaik and Xiaohui Tao and Yan Li and Christopher Dann and Jacquie McDonald and Petrea Redmond and Linda Galligan},
  title = {{A Review of the Trends and Challenges in Adopting Natural Language Processing Methods for Education Feedback Analysis}},
  booktitle = {IEEE Access},
  year = {2022},
  url = {https://openreview.net/forum?id=R1unjGbbtV},
}
Figure