AN AUTOMATIC FRAMEWORK FOR DOCUMENT SPAM DETECTION USING ENHANCED CONTEXT FEATURE MATCHING

Yenuga Padma, Dr.Y.K.Sundara Krishna

Abstract


With the growth in the communication systems, opinions became the most used communication method in the corporates, research and education. Nevertheless, with the increasing popularity the challenge for all internet service providers is to keep matching the demand for bandwidth. The major challenge to keep the bandwidth up to the usage is dealing with the spam messages. A spam communication or review is something that the sender uses for promotion and for the received may be useless. Thus for the receiver the messages are mostly unimportant. The detection of the spam reviews cannot be done at the review server end and need to done at the receiver side. Failing in detecting the spam can easily overload the review communication channel and reduce the effective use of the bandwidth. A number of researchers are carried out in order to detect the spam messages by deploying the filters. The outcomes are partially satisfactory as most of the parallel researches have demonstrated the rejection of the documents based on the pre-defined keywords. Nonetheless, these methods are not satisfactory as the use of words for every review writer may vary. As a result influenced by certain keywords, the receiver may lose some important communications. Thus the demand of the modern research is to enhance the detection of the spam reviews by using enhanced techniques rather than only depending on the keywords. This work proposes a novel automated framework powered by machine learning technique to detect the keywords and improve the detection by deploying context detection methods. The major outcome of this work is to build and demonstrate an automated framework for review spam detection with review rejection filters. The work outcomes into a highly satisfactory detection rate and demonstrate a sustainable model.

Keywords


Spam Detection, Opinion Detection, Spam Filtration, Machine Learning, Feature Detection

Full Text:

PDF


DOI: https://doi.org/10.26483/ijarcs.v9i1.5473

Refbacks

  • There are currently no refbacks.




Copyright (c) 2018 International Journal of Advanced Research in Computer Science