We recognize adversaries as fast-evolving hidden entities that can only be defeated by an information advantage promptly delivered to effectors.
To achieve information dominance, we observe the hostile digital environment as a multi-layered and multi-dimensional space. Meet XRATOR HyperCube.
Powered by our in-house R&D, we focus on delivering timely and accurate information for the platform users.
To provide accurate scoring, recommendation and course of actions for Proactive Risk Mitigation, XRATOR aggregates four main categories of information to model an organization: its business environment, its internal and external attack surface, the active relevant threats and its security posture.
XRATOR’s HyperCube unique capabilities reproduce the aggregation and processing of information by a team of risk expert.
XRATOR Hypercube ingest more data, compute more scenarios and deliver faster than any human can do.
XRATOR cloud-agnostic and pipeline-oriented architecture save semi-structured data provided by the internal and external scanners, third-parties API. It gathered unstructured data from social media, blog, news media, and research paper. These data are then cleaned, structured, normalized and organized.
Our automated Data Science Lab performs various convolution and scenarios until meeting quality criteria, providing the curated and updated HyperCube access through the Web Platform features.
Easy cloud deployment
The Hypercube data crunching machine is transparent for the organization, which only has to ensure the connectivity with its Private Cloud and place its internal scanners into its network according to its preferences.
Natural Language Processing
Cyber Risk and Cyber Security
Whatever our role in the Cyber Risk Management process, every disciplines must monitor news information impacting their fields to stay fit. There is no exception for cybersecurity and there is also an additional challenge : our adversaries evolves constantly. It is hard to stay on track of the myriads of cyber-attacks, malware evolution and cyber gangs activities. It is impossible to remember every tiny piece of information that may be crucial in a near or far future. It is challenging to keep an eye on all technical, social, geopolitical trend. That is where Natural Language Processing (NLP) enter in the game, by automating the security watch and the report production.
What is NLP ?
Human languages can be processed through Natural Language Processing (NLP), an interdisciplinary domain composed of Computer Science, Artificial Intelligence, and Linguistics studies. Machines use NLP to comprehend, analyze, manipulate, and interpret human language. Human can leverage Natural Language Understanding (NLU) to organize knowledge and perform tasks such as entity recognition, relationship extraction, and topic segmentation. In addition they can also employs the machine for Natural Language Generation (NLG) such as text summarization, translation, building virtual assistant or report generation automation.
How does it works?
For NLU, the machine first cuts the text into sentences, then words, and finally figures out how the words refer to one another in the sentence with grammar rules. It uses a dictionary to link words to their meanings. Once the machine understands the sentence’s meaning, it can put the text in context and reinterpret it. The inverse path, from interpretation to word construct, allows the machine to write a new text or sentence (NLG).
from XRATOR R&D Team
If you are curious about how XRATOR R&D Team operationalize Natural Language Processing for cyber prevention objectives, check out three of the projects they are continuously improving.
Cybersecurity Database Consolidation
Technical experts may know or use public databases such as MITRE CVE (Vulnerability), CWE (Weakness), CAPEC (Attack Pattern) or ATT&CK® (Attack Techniques).
Our team consolidate all these database, correct and enrich data entries and link them together to get a full normative view from the adversaries behavior to the organizations technical flaws.
Structured Cyber News Feeds
Drawing public research published online, our Unstructured Text Pipeline clean texts, spots attacker groups, malwares, vulnerabilities, modus operandi or victims mentions.
Using Named Entity Recognition (NER) and Relationship Extraction, the content is stored in a structured graph using the open-source Structured Threat Information eXchange (STIX) language.
Multiple problems, same solution
Vulnerability remediation are generally off-context, causing great efforts to technical expert. A vulnerability "X" remediation maybe be to update to version 1. But it was later discover that the version 1 is also vulnerable and is patched in version 2.
The R&D Team has developed algorithms to merge problems that share a similar final solution.