With Artificial Intelligence, Spam Content May Be History

It’s a well-known fact that product and service reviews and comments affect sales today. Around 71% of consumers say that buying a product works better after reading other people’s impressions, while 88% say reviews influence their purchasing decisions. That’s why serious work is being done to reduce spam and misleading reviews written by malicious third parties. Scientists at the Hartman Group and the University of Washington have made great progress in this challenging area.

744669abf6f2d078a35b5afa4b9d6c1b5ec099f2

In an article published on Arxiv.org, researchers described “spamGAN” (GAN for Semi-Controlled Opinion Spam Detection). GAN is the name given to the class of productive reverse networks in machine learning system.

The authors of the article, “Unwanted reviews and comments; It’s a common problem on e-commerce sites, social media, travel sites, and movie review sites. “We think it is necessary to define spam reviews as a classification problem and be classified as ‘spam’ or ‘non-spam’ when a review is given,” they said.

5df8ee2a4a79a995bb45ee76e59209bebe9575e6

In this method, unlabeled ingredients are used with a small amount of labeled content. In this way, machine learning is improved and this technique is called ‘supervised learning’.

The authors of the paper write, “Most of the current research on spam reviews (other than deep learning methods) uses experiential approaches to classify and describe spam behavior. However, in our GAN-based approach, the features are learned by the neural network. At the same time, we believe that SpamGAN can also generate spam and non-spam content that can be used for artificial data generation in unconfirmed situations.”

In addition, spamGAN will provide datasets and a more sophisticated classifier for use in future studies and experiments.

Related Posts

Leave a Reply

Your email address will not be published.