A review on predicting the occurrence of illegal gatherings using a set of machine learning methods

Nowadays, social networks have become a place of governance and a suitable platform for the enemies of the country. During the gatherings that have sometimes occurred in the country, it has been rightly proven that social networks and foreign messengers have been a place for organizing, managing, inciting, persuading and even training young people for rioting and vandalism. At present, with the rapid increase of all types of crimes, the traditional methods of crime investigation have not been able to provide the desired results, because their speed is slow and inefficient. The purpose of this study is how machine learning can be used by security or law enforcement agencies to detect, prevent and deal with illegal gatherings with a very accurate and fast speed. In order to achieve this goal, 73 articles in the period from 2012 to 2023 were analyzed in which machine learning methods were used. We can see that most of the articles use the machine learning approach with supervised and labeled data. However, artificial neural networks with 44%, random forest methods with 30% and K-nearest neighbor method with 26% were the most commonly used methods. Also, 62% of the researchers have used criminal data sets from public portals on the Internet and 38% have used official and private data sets of legal organizations, including the police, in their research. The results of the research show that the use of random forest methods has the best performance, but for large data sets, the use of artificial neural network methods has provided the best results for predicting the occurrence of crime based on its time and place.