Content-based approaches to detecting mobile spam to date have focused mainly on analyzing the topical aspect of a SMS message (what it is about) but not on the stylistic aspect (how it is written). In this paper, as a preliminary step, we investigate the utility of commonly used stylistic features based on shallow linguistic analysis for learning mobile spam filters. Experimental results show that the use of stylistic information is potentially effective for enhancing the performance of the mobile spam filters. spam classifiers with additional features, such as orthogonal sparse word bi-grams (Cormack et al., 2007a; Cormack et al.,.
