Twitter Spam has become an essential drawback these days. Recent works specialize in applying machine learning techniques for Twitter spam detection that build use of the applied math options of tweets. In our tagged tweets dataset, however, we tend to observe that the applied math properties of spam tweets vary over time, and therefore the performance of existing machine learning based classifiers decreases. This issue is referred to as “Twitter Spam Drift”. In order to tackle this problem, we firstly carry out a deep analysis on the statistical features of one million spam tweets and one million non-spam tweets, and then propose a novel Lfun scheme. The projected scheme can discover “changed” spam tweets from unlabelled tweets and incorporate them into classifier’s training process. a number of experiments are performed to evaluate the proposed scheme. The results show that our proposed Lfun scheme can significantly improve the spam detection accuracy in real-world scenarios.