How Could Artificial Intelligence Help to Make a more Equitable Digital Society?

Rafael Benítez Rochel, Francisco L. Valverde

The desire and the need to communicate has been present in humanity from the beginning of the species of the human being. Undoubtedly, the impact and influence of communication, and in particular the broadcasting, has been and continues to be a factor of the first magnitude in the evolution of society. The rise of the Internet and especially the web 2.0 phenomenon, where anyone has the possibility of broadcasting, has been very beneficial for society. However, a malicious use of this technology can cause unfair damage to a part of society to achieve an economic or political objective. But the good news is that artificial intelligence can help correct this malicious use of information technology. Our research focuses on systems based on artificial intelligence to detect false content in web 2.0


Social media are web 2.0 applications, websites, forums, social networks such as Facebook, Instagram and others, blogs, etc. This can be anything shared by people over the Internet and created by people for other people online. The desire to use social media is hidden in human nature. This is the main reason for the remarkable growth of social networks in recent years. We can find certain advantages of social media for society such as connectivity (we can share our thoughts), education (it is very easy to educate others who are experts and professionals), Real Time News and Updates, etc. However, it has also affected the society in the negative way. It is in the hands some users to use the advantages mentioned above to have, willingly or unwillingly negative impacts on the rest of users (Cyberbullying, Hacking, Addiction, Fraud and Scams …)

The negative and positive effects of social networks on companies could be expressed as increasing brand loyalty/affinity, reducing the cost of acquisition and scale, building digital brand equity in the positive side and Fake Reviews (deceptive opinions, spam opinions, or spam reviews) in the other one. The last problem it is an important issue because, nowadays, it is very easy and fast to generate fake reviews with algorithms. So, the detection of this kind of spam is a well-recognized problem that has attracted significant interest from the research community (Mukherjee et al., 2011),(Wang et al., 2011).

To clarify the above concern, many traditional machine learning-based methods have been employed, such as support vector machines (SVM) (Silva, et al 2013), (Alberto et al 2015) , (Silva et al 2017), naïve Bayes (Silva et al 2013), (Alberto et al 2015) , (Silva et al, 2017), decision trees (DT) (Silva et al, 2013), (Alberto et al 2015) , , (Silva et al, 2017), (Najada et al, 2014, (Alsaleh et al, 2015), and k-nearest neighbors (KNN) (Silva, Almeida et al, 2013), (Alberto et al 2015) , Random Forest  (Breiman, 2001) and Logistic Regression (Cox, 1958).

It is possible to classify a review as probably authentic or probably false using only a Support Vector Machine. But it is important to notice that computer-generated fake reviews continue to become increasingly sophisticated and difficult to detect. Hence, algorithms should be adjusted to detect deceptions created by artificial intelligence. Finally, no detection tool can guarantee 100% accuracy. Our research has focused on the detection of false reviews using artificial intelligence and, as a result, we present a method based on the Named Entity Recognition to simplify the analysis of large quantities of a wide variety of reviews. Only those reviews that have been classified as false are analyzed manually by a team of experts to ensure and confirm this classification. We defend that using a combination of different techniques and methods improves accuracy.


A new hybrid method to detect false reviews has been presented in this article. The main contribution is the simplicity that makes it suitable for analyzing large amounts of data. It is a decision support system for companies specialized in the detection of false news since it facilitates their work and greatly reduces the accuracy, speed and costs.