Reading product reviews before buying the product becomes a habit, especially for potential
customers. If they want to buy a product, they usually read reviews from some customers
about the current product. If the review is mostly positive, there is a big chance to buy the
product, otherwise if it’s mostly negative, they tend to buy other products. While, for a
company, the positive reviews from customers can generate significant financial benefits
for businesses, it can be used as input for decisions related to product design and what
services are provided to customers (Heidary et al., 2015). Related with the financial benefits
gained as a result of the positive reviews about the product / service from the customer, the
fraudsters tried to play the existing system by writing fake reviews and provide an
assessment that is not fair to promote or discredit a product or service (Heidary et al., 2015).
Individuals like this are called spammers and their activity called opinion spamming.

Automatic detection of spammers is a very important work but still lacks in research. Unlike
other types of spam, such as web spam or email spam, spam on a review is far more difficult
to detect. The main reason is that spammers can easily disguise themselves. Thus it is
difficult, for users to recognize, while web spam or email spam, one can determine spam or
not without much difficulty (Heidary et al., 2015). The other reason is difficult to find good
data (gold standard) related with fake and genuine reviews to build a model, because it is
very difficult, to manually identify/labeling whether the review is fake or not just by reading
them (India, 2012)