Prediction Retweet Using User-Based, Content-Based and Time-Based Features with ANN-Firefly Classification Method

Alif Aqshal (Faculty of Exact Sciences Education, Universitas Pendidikan Muhammadiyah Sorong, Papua, Indonesia)
Indri Anugrah Ramadhani (Faculty of Exact Sciences Education, Universitas Pendidikan Muhammadiyah Sorong, Papua, Indonesia)

ARMADA : Jurnal Penelitian Multidisiplin
Vol. 4 No. 5 (2026): ARMADA : Jurnal Penelitian Multidisplin, Mei 2026

Retweets on the Twitter platform serve as a critical indicator of information dissemination. However, there is a lack of effective predictive systems to estimate whether a tweet will be retweeted. This research aims to develop a retweet prediction model by leveraging a combination of user-based, content-based, and time-based features. The model is constructed using an Artificial Neural Network (ANN) and optimized through the Firefly Algorithm (FA) to enhance classification accuracy. The dataset was collected from Twitter using a data crawling technique via the Tweepy library, focused on the keyword “Covid Vaccination,” resulting in 12,796 Indonesian-language tweets. The collected data underwent preprocessing stages, including text cleaning, tokenization, normalization, and stemming. The ANN-FA model was then trained to classify the likelihood of a tweet being retweeted. Experimental results demonstrate that the ANN-FA model achieved an accuracy of 90.29%, outperforming the baseline ANN model without optimization. These findings indicate that applying the Firefly Algorithm significantly improves classification performance. The contribution of this study lies in developing a retweet prediction system that integrates multi-dimensional features with metaheuristic optimization, which can be utilized to support digital information dissemination strategies on social media platforms