Hoax Information Detection System Using Apriori Algorithm and Random Forest Algorithm in Twitter
Mailia Putri Utami, Oky Dwi Nurhayati, Budi Warsito
Abstract
This research is based on the disturbance faced by Twitter users, related to the distribution of hoax information in the text form. One of the efforts to overcome such problem is by building a system to detect hoax news in Twitter application. A set of information in text from on social media, can capture the use of language in written or verbal (corpus) form. Based on the advantages of Apriori algorithm which it is able to mine the text data from many used datasets and able to find the relation pattern or itemset combination in a database as a recommendation of the raised pattern. Moreover, the detection system also needs a method to classifying information based on the classes of hoax and non-hoax. One of the algorithms used is Random Forest algorithm, which it is able to combine several models of decision trees to eliminate the problems of overfitting. The purpose of this research, apart from implementing and integrating the Apriori algorithm and the Random Forest algorithm, is to make it easier for researchers to analyze and evaluate the system's results to detect hoax information most optimal level of accuracy. The results show that the system can detect hoax and non-hoax news, whose data is integrated directly with the Twitter application. Accuracy level, precision and recall from the built system in detecting hoax news information reached 100% with minimum support value of 23.