IP Library Granted Patent US 11,520,900
Granted Patent B2
US 11,520,900 · App. 16/548,329 · Granted Dec 6, 2022

Systems and methods for a text mining approach for predicting exploitation of vulnerabilities

Inventors: Nazgol Tavabi (Marina Del Rey, CA); Palash Goyal (Los Angeles, CA); Kristina Lerman (Los Angeles, CA); Mohammed Almukaynizi (Chandler, AZ); Paulo Shakarian (Chandler, AZ)
Assignees: Arizona Board of Regents on Behalf of Arizona State University; University of Southern California
G06F21/577G06F40/279H04L51/216H04L63/1416H04L63/1425H04L63/1433G06F2221/033
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 11,520,900
App. No.
16/548,329
Granted
Dec 6, 2022
Kind
B2
Abstract

Various embodiments of a computer-implemented framework for predicting exploitation of software vulnerabilities are disclosed.

Claims (7)

1. A method, comprising:

utilizing a neural network to learn a context-based distributed representation from data associated with the deep or dark web, the data including discussions about vulnerabilities and corresponding exploits; and

training a classifier utilizing the context-based distributed representation as a feature to the classifier such that the classifier is configured to output a classification for predicting an exploitation of a vulnerability associated with a communication, the classification considering a context of words from the communication,

wherein the classifier is implemented using a support vector machine with a radial basis kernel, wherein the support vector machine classifies communications by finding a set of hyper-planes that best separates each of the communications into a class.

2. The method of claim 1 , wherein the context-based distributed representation includes a paragraph embedding defined by a distributed representation of an entire post of the data to learn a global context of words in the entire post.

3. The method of claim 2 , wherein the context-based distributed representation comprises a word embedding that projects words in a lower-dimensional vector space with d dimensions, so that each word w i is represented by a d-dimensional vector.

4. The method of claim 1 , wherein the classifier utilizes further features including a frequency of mention or a common vulnerability scoring system (CVSS) score.

Assignments (3)
CONFIRMATORY LICENSE Recorded Jul 13, 2022
From: UNIVERSITY OF SOUTHERN CALIFORNIA
To: GOVERNMENT OF THE UNITED STATES AS REPRESENTED BY THE SECRETARY OF THE AIR FORCE
Reel/Frame 060649/0931 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jan 16, 2020
From: TAVABI, NAZGOL; GOYAL, PALASH; LERMAN, KRISTINA
To: UNIVERSITY OF SOUTHERN CALIFORNIA
Reel/Frame 051538/0486 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 3, 2019
From: ALMUKAYNIZI, MOHAMMED; SHAKARIAN, PAULO
To: ARIZONA BOARD OF REGENTS ON BEHALF OF ARIZONA STATE UNIVERSITY
Reel/Frame 050247/0309 →
Continuity (2)
Provisional Application 62721401 · Aug 22, 2018
Related Publication 20220229912A1 · Jul 21, 2022
Cited By (1)
US 12,436,827