Many researchers were attending the conferences, where we could met researchers in AI, in complex systems, on Blockchain, and many other topics related to Decision Making.
The opening speech was focused on fraud detection. The speaker, V.S. Subrahmanian, Dartmouth College Distinguished Professor in Cybersecurity, Technology, and Society, presents some of his work on detection of fraudulent posts influencing election campaigns, or forged opinions, posted by products retailers themselves, or their competitors.
As you may know, products are evaluated with stars, usually within a scale from 1 to 5. One star weight around 5 to 10 percent of turnover. The battle for the stars then really becomes a major issue in e-commerce. Thanks to research in Artificial Intelligence, it is possible to modelize the relationship between products, their rating, shoppers, their fairness, taking into account, the number of advices, the elapsed time between the purchase and the advice, etc…Calculations made on the various e-commerce platforms allow to deduce which shoppers are not honest, and allow these platforms to delete fraudulent advices or block fraudulent accounts. This has become a full time task in all the big e-commerce platforms.
The same kind of approach has been set up for posts in election campaigns. Artificial Intelligence enables to detect fraudulent post or comments, on platforms like Facebook or Twitter.
At ChainHero we presented our work on “Fuzzy Rules Based Solution for System Administration Security Management via a Blockchain”. Blockaudit, our product, gathers kernel events from sensitive servers (more informations here : https://chainhero.io/blockchain-products/blockchain-audit/).
Among all these events, we need to detect events corresponding to a rogue behavior and generate alerts for security teams. The fuzzy logic helps to create more flexible rules to detect abnormal behaviors. We use a description specific language with keywords to qualify amounts, for example HAPPEN, FEW, MODERATE, MANY. The fuzzy logic will translate these keywords, to values depending on the number of passed events. So MANY, will not be the same amount if there are less than 100 events during the night, or more than 10 000 during working hours !
The paper describing this work on Fuzzy Logic is available on the Blockchain Congress website https://link.springer.com/chapter/10.1007/978-3-030-23813-1_9
By the way, the location of the Blockchain Congress, Ávila city, is a very nice place with impressive walls, which are classified at the Unesco. The “chuleton”, the veal chops, were not bad either !
Another subject, not related to mathematics, has been presented : The Electronic Bill of Lading, which involve more legal and habits issues. In Germany, new laws for paperless administration, try to change the habits for Bill of Lading, where paper was the norm.
The Bill of Lading proves that goods were taken over in the described form by a carrier. You can easily imagine the thickness of the document, which is stamped, signed by all the different parties, which interact during the containers travel. The Blockchain is well suited in order to store all these informations and transactions.
The challenge, here, is to change habits, which have been used for centuries. But isn’t it the case with all disruptive technologies to fight to transform old habits?