Twitter Spam Detection Using Machine Learning
Freddy Jenson, Grace Joseph · Zenodo (CERN European Organization for Nuclear Research) · 2022
Abstract— Spam on Twitter has become a common occurrence these days. The most recent work focuses on using machine learning techniques to detect twitter spam using factual parts of tweets. We observe that the measurable attributes of spam tweets fluctuate over time in our labeled tweets informative collection, and that, as a result, the existing machine learning-based classifiers' effectiveness is deteriorating. effectiveness of existing machine learning-based classifiers degrades. To address this issue, a strategy known as the Lfun approach can be used, that can detect modified spam twitter tweets from unlabeled tweets and merge them into the process of classifier training. The updated data that is trained is used to generate a fresh dataset with unlabeled tweets, resulting in the detection of tweets that are spam. The proposed technique can alter training data, such as deleting old observations after a specific period of time and to save space by removing useless data.