Abstract

AbstractText classification is a basic task in the field of natural language processing, and it is a basic technology for information retrieval, questioning and answering system, emotion analysis and other advanced tasks. It is one of the earliest application of machine learning algorithm, and has achieved good results. In this paper, we made a review of the traditional and state-of-the-art machine learning algorithms for text classification, such as Naive Bayes, Supporting Vector Machine, Decision Tree, K Nearest Neighbor, Random Forest and neural networks. Then, we discussed the advantages and disadvantages of all kinds of machine learning algorithms in depth. Finally, we made a summary that neural networks and deep learning will become the main research topic in the future.

Highlights

  • Machine learning mainly studies how to learn unknown rules from given data, that is, to find some objective rules from observed data, and use the learned rules to analyze and predict irregular data or unknown events

  • Text classification is a basic task in the field of natural language processing

  • Text classification algorithm based on machine learning is usually divided into four steps: 1) Features extracting and document modeling

Read more

Summary

Introduction

Machine learning mainly studies how to learn unknown rules from given data, that is, to find some objective rules from observed data (samples), and use the learned rules (models) to analyze and predict irregular data or unknown events. The main machine learning method is to conduct statistical analysis on the existing marked data, by which we can find laws and obtain models. Using these trained models, we can make prediction and analysis on unknown data to obtain classification results. Text classification algorithm based on machine learning is usually divided into four steps: 1) Features extracting and document modeling. In order to solve these problems, features extracted by traditional methods are usually dimensionality reduced, and a subset of features that can best represent text information and achieve the best classification effect are selected among all features. The accuracy rate refers to the proportion of samples with positive classification results that are truly positive; the recall rate refers to how many samples of all positive classes can be correctly classified by the classifier; the measurement value of F is the geometric average of the accuracy rate and recall rate

Principles of Machine Learning Algorithms
Naive Bayes
Supporting Vector Machine (SVM)
Decision Tree
KNN (K-Nearest Neighbor)
Random Forest
Neural Network
Comparative Study of the Machine Learning Algorithms
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call