Check out Threat Flow, the Security Industry’s First Transparent Generative AI Application

Bsides 2021 – Flare booth and conference

Bsides Montreal 2021

CONFERENCE
Automatic Extraction of content from criminal underground forums.

The speaker

Olivier Michaud

Olivier Michaud 

AI Expert @Flare Systems

Olivier is currently a master’s student in software engineering with a concentration in artificial intelligence at the École de technologie supérieure (ETS). His achievements allowed him to start this master’s degree in his last year of a bachelor’s degree in software engineering at the same school. Proud representative of Quebec at the Canadian Competition of Engineering in 2020, Olivier distinguished himself by winning the excellence scholarship from the École de technologie supérieure in order to continue his studies at the graduate level. His interest for artificial intelligence has led him to work with Flare Systems, now a partner in his research. During his undergraduate studies, Olivier was particularly involved with the Lan ETS Club in addition to being tasks with running laboratory classes.

Abstract of the conference

Automatic Extraction of content from criminal underground forums.

September 25, 2021 @10:45 AM EST

With the advent of new communication channels that take advantage of anonymity technologies and cryptocurrencies, the criminal underground has grown significantly over the past 10 to 20 years. Malicious actors within this underground use illicit markets and discussion forums to communicate on and transact increasingly advanced techniques and tools to extort, steal identities and data. Many faclitating platforms are hosted on the Tor network, a communication channel that is part of the dark web. The anonymity that the Tor network provides has made it increasingly difficult for law enforcement agencies and insttutions impacted by these attacks to protect themselves. According to a survey carried out by the CyberEdge Group, 86% of companies with more than 500 employees were victims of a successful cyberattack in 2020. In addition, 60% of non-drug related advertisements for illicit goods and services posted on the dark web were deemed likely to impact a business (Guccione , 2021). Monitoring the criminal underground helps potential victims detect past and future attacks, and put up more effective responses to them. In other words, it helps victims develop a proactive strategy to better face today’s threats. In order to automate the process of monitoring the criminal underground, technologies from the field of web Crawling and web scraping have been proposed. Web crawling refers to the process of browsing the web automatically using a robot to index its content. Web scraping, on the other hand, is the set of techniques used to extract content from an HTML page. These technologies usually rely on manual configurations, which further increases operating costs. In this conference, we propose a method to automate data collection of forums in the criminal underground. More specifically, the focus of this presentation is on pages containing forum topics, from which the title, author and publication date of each entry will be extracted. The proposed method makes it possible to transform an HTML page in order to carry out “sequence labeling”, a technique in the field of natural language processing. This not only works on forums used during training, but can also be generalized to other unknown forums.

Download the full program

Meet Flare there!

Our team will be at Bsides all day September 25th, 2021.


Get Your Tickets

Share This Article

Eric Clay

Related Content