NeuraSecure
Secure whatever,whenever,wherever, from the latest threats
Created on 10th November 2024
•
NeuraSecure
Secure whatever,whenever,wherever, from the latest threats
The problem NeuraSecure solves
In 2023, organizations faced an average of 1,158 cyber incidents per week, a 1% increase from the previous year. Ransomware attacks surged by 42%, with over 5,600 incidents reported. Our solution addresses these challenges by providing real-time feeds of cyber incidents targeting critical sectors like finance, healthcare, power, and public services.
The platform offers real-time monitoring and analysis of cyber incidents, enabling stakeholders to respond quickly to mitigate risks. It manages the entire process, from data collection to analysis and reporting, integrating web crawling, machine learning, and data visualization to ensure accurate, up-to-date information.
We simplify cybersecurity by offering helpful tips and visuals, making complex concepts easier to understand. A chatbot explains basic cybersecurity terms and risks to further support users' knowledge.
Users can customize their experience by subscribing to keywords or technologies they use, such as SQL. If a threat related to their selected keywords is detected, they receive email notifications with details and protective measures, such as upgrading to a safer version(real time alert). This ensures users stay informed about relevant threats and can take quick action to protect their systems.
Features and Option
-> Threat Protection and Solution
-> Online Risk Management
-> Realtime Alert
-> Deep Insights
-> Data Visualization
-> Check are you Pwned with latest attack
-> Bug Bounty Hunter
Challenges we ran into
Challenge 1: Extracting Relevant Cyber Incident Data from Large Volumes of Information
Isolating accurate cyber incident data from a vast pool of online content was a key challenge. The task required filtering through noise while ensuring timely, relevant information about incidents in Indian cyberspace.
Solution:
We designed a two-step filtering system. First, a batch crawler seeded with trusted sites searched for specific HTML attributes, such as time tags, to detect recent cyber events. The crawler used keywords (e.g., "cybersecurity," "data breach") to decide whether to follow a URL. Non-relevant links were discarded immediately, optimizing resource use and improving accuracy. The relevant URLs were then saved in a CSV file for the next step.
Challenge 2: Ensuring Cyber Incident Relevance with High Accuracy
After gathering URLs, we needed a way to confirm that each URL’s content genuinely pertained to Indian cyber incidents. Accurately filtering for relevance required automated yet precise content validation.
Solution:
We trained a Naive Bayes-based machine learning model with historical data related to cyber incidents, specifically in India. This model leveraged keywords associated with cyber incidents and tested them to enhance its detection accuracy. After gathering URLs, the crawler passed them to the ML model, which then performed content verification on the extracted data, ensuring only relevant cyber incident information was retained.
Keyword Strategy for Effective Coverage
Our filtering included a diverse keyword set for comprehensive coverage: general terms (e.g., "cyber attack," "phishing"), sector-specific terms (e.g., "banking data breach," "government hack"), and technical terms like "SQL injection" and "advanced persistent threat." This strategy maximized relevant data capture and enhanced filtering precision.