Analysis of child development facts and myths using text mining techniques and classification models

Area of Science:

• Computational Linguistics
• Child Development Studies
• Artificial Intelligence

Background:

• Online misinformation regarding child development poses risks to children's well-being.
• Existing research has not sufficiently addressed the distinction between child development myths and facts using computational methods.

Purpose of the Study:

• To develop and evaluate text mining and machine learning models for classifying child development information as myth or fact.
• To provide parents and caregivers with a tool to identify reliable child development resources.

Main Methods:

• Data collection from publicly available websites on child development.
• Application of text mining for data pre-processing.
• Evaluation of six Machine Learning (ML) classifiers and one Deep Learning (DL) model using two feature extraction techniques (Bag-of-Words) and cross-validation (k-fold, leave-one-out).

Main Results:

• Logistic Regression (LR) with Bag-of-Words (BoW) achieved the highest classification accuracy at 90%.
• LR demonstrated exceptional speed and efficiency, with a testing time of 0.97 μs per statement.
• The model effectively distinguished between factual and mythical child development information.

Conclusions:

• Logistic Regression combined with Bag-of-Words is a highly accurate and efficient method for classifying child development information.
• This approach can serve as a valuable tool to combat misinformation and support informed decision-making for parents.
• Further research can expand this methodology to other areas of health and parenting information.