A hybrid spam detection method based on unstructured datasets

Abstract

Abstract:The identification of non-genuine or malicious messages poses a variety of challenges due to the continuous changes in the techniques utilised by cyber-criminals. In this article, we propose a hybrid detection method based on a combination of image and text spam recognition techniques. In particular, the former is based on sparse representation based classification, which focuses on the global and local image features, and a dictionary learning technique to achieve a spam and a ham subdictionary. On the other hand, the textual analysis is based on semantic properties of documents to assess the level of maliciousness. More specifically, we are able to distinguish between meta-spam and real spam. Experimental results show the accuracy and potential of our approach. Abstract The identification of non-genuine or malicious messages poses a variety of challenges due to the continuous changes in the techniques utilised by cyber-criminals.In this article, we propose a hybrid detection method based on a combination of image and text spam recognition techniques. In particular, the former is based on sparse representation based classification, which focuses on the global and local image features, and a dictionary learning technique to achieve a spam and a ham sub-dictionary. On the other hand, the textual analysis is based on semantic properties of documents to assess the level of maliciousness. More specifically, we are able to distinguish between meta-spam and real spam. Experimental results show the accuracy and potential of our approach.

A Non-genuine Message Detection Method Based on Unstructured Datasets

“…Suppose that G can be embedded onto two different hyperplanes H 1 and H 2 , such that the former is the semantic layer, and the latter is the non-genuine layer, which contains the concepts related to spam messages and their mutual relationships. For full details, please refer to [1]. In particular, H 2 contains sub-networks describing concepts which might be part of a non-genuine message, as well as their corresponding synonyms.…”

“…In particular, semantic networks can be utilised to identify and assess the level of maliciousness of a message. As defined in [1], assume that…”

“…[2]. However, there is a compelling argument regarding the concept of non-genuine communication [1]. Spam emails are certainly part of non-genuine communications, where a user receives unsolicited emails on a variety of topics.…”

“…i = i + 1 10: end while 11: return X Algorithm 1 incrementally updates the network by considering newly available textual datasets [1], and each time a new text is added the steps above are carried out updating the network by merging it with new nodes and edges. The process of merging the X i 's into X is carried out recursively, so that identical nodes and edges are merged, as there is no loss of information.…”

“…In fact, the relation rel between NP1 and NP2 has general semantic connotations, and so, there is no need to diversify them. A seed is defined as a sub-network S ⊂ G [1]. The level of non-genuineness of S is then assessed depending on the number of components of S ∩ H 2 , i.e.…”