Twenty Years of Machine-Learning-Based Text Classification: A Systematic Review

Abstract

Machine-learning-based text classification is one of the leading research areas and has a wide range of applications, which include spam detection, hate speech identification, reviews, rating summarization, sentiment analysis, and topic modelling. Widely used machine-learning-based research differs in terms of the datasets, training methods, performance evaluation, and comparison methods used. In this paper, we surveyed 224 papers published between 2003 and 2022 that employed machine learning for text classification. The Preferred Reporting Items for Systematic Reviews (PRISMA) statement is used as the guidelines for the systematic review process. The comprehensive differences in the literature are analyzed in terms of six aspects: datasets, machine learning models, best accuracy, performance evaluation metrics, training and testing splitting methods, and comparisons among machine learning models. Furthermore, we highlight the limitations and research gaps in the literature. Although the research works included in the survey perform well in terms of text classification, improvement is required in many areas. We believe that this survey paper will be useful for researchers in the field of text classification.

Enhancing ASD detection accuracy: a combined approach of machine learning and deep learning models with natural language processing

“…However from the whole number of models that were tested, the best one was the DL model called BERTweet with an accuracy of 87.7%. This aligns with previous studies that have shown that BERT-based models have a great effectiveness in categorizing any kind of text texts, including tweets [18,31,32].…”

Ensemble Methods to Optimize Automated Text Classification in Avatar Therapy

“…It is also to be noted that the literature reviews on ensemble methods highlight that ensemble modeling is an acceptable technique for coping with individual classifiers' large variation while minimizing general mistakes [49]. Furthermore, ensemble techniques are reported to be an appropriate method to improve accuracy in text classification tasks, which is what has been observed in its use in AT [50]. Interestingly, a recent study comparing the use of single classifier to an ensemble approach in the domain of mental health suggests that for the prediction of mental health problems, ensemble models demonstrate better prediction results [51].…”

Using automated text classification to explore uncertainty in NICE appraisals for drugs for rare diseases

“…In this study, we manually coded paragraphs from NICE appraisals for RDTs to train and test three models: Naïve Bayes (21), Lasso regression (22), and Support Vector Machines (SVM) (23). These are frequently used models for text classification, tend to have good classification performance, and are simpler and computationally cheaper to implement than some more sophisticated supervised learning approaches (24–27). We used the caret (28), glmnet (29), e1017 (30), and quanteda.textmodels (31) packages to estimate the models.…”