Systematic review of content analysis algorithms based on deep neural networks.

Area of Science:

• Computer Science
• Data Science
• Artificial Intelligence

Background:

• The exponential growth of data necessitates advanced methods for knowledge extraction.
• Text mining is crucial for deriving insights from unstructured textual data.
• Information technology advancements accelerate data flow, increasing data warehousing demands.

Purpose of the Study:

• To systematically review and analyze deep learning and machine learning algorithms used in text mining.
• To identify the most effective techniques for classification and knowledge extraction from text data.
• To compare various neural network architectures including ANN, RNN, CNN, and LSTM.

Main Methods:

• Conducted a Systematic Literature Review of studies from 1997 to 2021.
• Retrieved 130 relevant studies from electronic databases.
• Selected 43 studies for in-depth analysis based on inclusion and exclusion criteria.

Main Results:

• Hybrid Long Short-Term Memory (LSTM) emerged as the most frequently used deep learning algorithm.
• Support Vector Machines (SVM) demonstrated high accuracy among machine learning methods.
• Analysis covered various neural networks like Artificial Neural Networks (ANN), Recurrent Neural Networks (RNN), and Convolutional Neural Networks (CNN).

Conclusions:

• Hybrid LSTM and SVM are highly effective for text mining tasks.
• The study provides insights into the landscape of algorithms used in text mining.
• Future research can build upon these findings for enhanced data analysis.