Network Security Database Filtering Tool Using Cosine Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network security systems fail to quickly and accurately identify and filter potential threats from user-submitted messages, leading to delayed detection and resolution of security issues, which can affect network functionality.
Innovation Solution
A network security database tool equipped with a translation engine and filtering engine that creates a word vector from user-submitted messages, calculates cosine differences with stored group vectors, and assigns messages to relevant threat groups based on these differences, enabling rapid identification and filtering of security threats.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If manual analysis of user-submitted messages is used to identify security threats, then accuracy of threat identification may be maintained, but the speed of detection and response is significantly reduced
Solution Approach 1:
The patent introduces cosine similarity as an intermediary mathematical measure to bridge the gap between automated processing and accurate threat identification. By calculating the cosine similarity between word vectors of submitted messages and pre-stored group vectors representing different threat categories, the system achieves both rapid automated classification and accurate threat identification without requiring manual analysis
Solution Approach 2:
The patent replaces the mechanical manual analysis process with an automated computational system that uses vector mathematics. Instead of human analysts manually examining messages, the system automatically converts messages into word vectors, calculates cosine similarities with stored group vectors, and rapidly classifies threats, thereby dramatically improving detection speed while maintaining accuracy
2Reliability
If all user-submitted messages are analyzed in detail, then comprehensive threat detection is achieved, but the time required for processing increases significantly
Solution Approach 1:
The patent applies partial action by using cosine similarity calculation as a efficient filtering mechanism. Instead of performing exhaustive detailed analysis on all messages, the system first computes the cosine similarity between message word vectors and group vectors to quickly identify and prioritize potentially malicious messages, thereby reducing overall processing time while maintaining comprehensive threat detection for prioritized messages
Solution Approach 2:
The patent segments the message analysis process into distinct stages: (1) converting messages to word vectors, (2) calculating cosine similarities with pre-stored group vectors, (3) prioritizing messages based on similarity scores, and (4) performing detailed analysis only on prioritized messages. This segmentation allows the system to efficiently process large volumes of messages while maintaining comprehensive detection capabilities
Data Source
AI summary
A method comprises creating a word vector from a message, wherein the word vector comprises creating a word vector from a message stored in a database, wherein the word vector comprises an entry for each word of the plurality of words and each entry indicating a number of occurrences in the message of the word for that entry. The method further comprising calculating a cosine difference between the word vector and each group vector of a plurality of group vectors and assigning the word vector to a group for a group vector of the plurality of group vectors based on the calculated cosine difference between the word vector and the group vector for the group. The method may also comprise reformatting the message to include a phrase and assigning the phrase to the same group that the word vector is assigned to. The method may also comprise notifying the user.