Research pick: Mowing the astroturfers - "Controlling astroturfing on the internet: a survey on detection techniques and research challenges"

A grassroots movement is one that emerges and evolves naturally, growing new support as it does so. “Astroturfing” is the opposite of that. It is a movement support for which is bought and paid for. It has the look of a grassroots movement, but a closer inspection reveals it to be fake. Now, writing in the International Journal of Web and Grid Services, Australian computer scientists have surveyed the techniques available to detect astroturfing on the internet. The term derives from the synthetic green grass – AstroTurf – often used in sports arenas and public areas as an alternative to living turf.

Syed Mahbub and colleagues at La Trobe University, in Melbourne, explain that astroturfing represents a significant threat in the business world, in politics, public health, and many other realms. Fake support for a controversial system, product, or service can persuade unwitting observers and stakeholders of merit, where no merit is due. This can have dire consequences for genuine political candidates in an election, for instance, or for sales of better rival products, and adoption of systems and services that are in reality better than the astroturfed ones. Political blogs, news portals, and review websites are carpeted with Astroturf to the detriment of everyone but the astroturfers and their associates.

At its most mundane, astroturfing might lead to someone buying a, perhaps inferior, green widget from company A in preference to the better blue widget from Company B. At the other extreme, one might see a politician achieve election success where support has been entirely faked and the electorate duped into disregarding the genuine candidate.

Researchers in social media, e-commerce, and politics, are looking to find detection methods for spotting astroturfing. Mahbub and colleagues point out that there are content analysis techniques, individual and group identification techniques, linguistic feature analysis, authorship attribution techniques, and machine learning all being used with varying degrees of success to detect astroturfing.

“Astroturfing, in the present world, is a global phenomenon,” the team writes. “The magnitude of its effect is significantly threatening the integrity and consistency of information we receive from the internet. Thus, the prevention and detection of astroturfing demand more attention from the research community.” Their research paper offers researchers a taxonomy of those detection techniques that might help in the development of better approaches to the detection of this insidious problem.

Mahbub, S., Pardede, E., Kayes, A.S.M. and Rahayu, W. (2019) ‘Controlling astroturfing on the internet: a survey on detection techniques and research challenges’, Int. J. Web and Grid Services, Vol. 15, No. 2, pp.139–158.