Veneta Yosifova

Dr. Veneta Yosifova

orcid

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 May 2021
Verified
Conference Paper
Closed access
VYVeneta Yosifova
Veneta Yosifova
ATAntoniya Tasheva
Antoniya Tasheva
RTRoumen Trifonov
Roumen Trifonov
Vulnerability type is not part of the standard CVE scheme so the ability to determine it only on the basis of text description would be a very useful for automated vulnerability handling. The growing number of hardware and software vulnerabilities discovered every year makes it more difficult for manual classification of the vulnerabilities types. This justifies the need for automatic machine learning classification. In this study we research the performance of base ML classifier algorithms, such as Linear Support Vector Classification, Naive Bayes, and Random Forest Classifier. To measure the performance of our classifiers, we use precision, recall, and f1-score evaluation metrics. Previous studies have focused on machine learning methods predicting platform vendor and products, vulnerability scoring, software vulnerabilities exploitation. Our study aims to show that machine learning is suitable for automated vulnerability type classification.

26 September 2019
Verified
Conference Paper
Open access
VYVeneta Yosifova
Veneta Yosifova
RTRoumen Trifonov
Roumen Trifonov
ATAntoniya Tasheva
Antoniya Tasheva
ONOgnian Nakov
Ognian Nakov

30 September 2021
Verified
Conference Paper
Closed access
VYVeneta Yosifova
Veneta Yosifova
A typical vulnerability database contains many records and every record has assigned a unique identifier and a description. Common Vulnerabilities and Exposures (CVE) database is one of the most common, free to use, large and reliable vulnerability databases. We provided this research due to the growing number of discovered vulnerabilities, roughly fifteen thousand for every year, which makes the manual classification very difficult. In this paper we prove that ensemble machine learning is an adequate technique for automated vulnerability type classification. We also achieved noticeably better prediction scores compared to basic machine learning methods, using improved training features and ensemble machine learning algorithms.

17 May 2024 in Preprints.org
Verified
Preprint
Open access
PDF file
VYVeneta Yosifova
Veneta Yosifova
This paper explores the role of open-source large language models in IoT cybersecurity world. The threats of malicious activity on the Internet and the loss of private information are very real and lead to serious consequences. The purpose of this paper is to investigate how open source-large language models can help to defend against the growing threat of cyber-crimes. We conducted our experiments in two directions. The first one is a security assistant that helps with cybersecurity best practices advices. The second one is a how large language model can simulate a vulnerable IoT system. For both types of experiments, the interactive mode of operation of the language model is used. In the context of the cybersecurity research, a major advantage of the locally installed open-sourced large language models is that they do not share sensitive data with a remote system in a cloud. The paper concludes by discussing the potential impact of open-source large language models on cybersecurity research and recommends future research directions.


6 October 2022
Verified
Conference Paper
Closed access
VYVeneta Yosifova
Veneta Yosifova
Incidents handling follows developed methodologies for dealing with incidents such as: cyber-attacks, breaches in the organization's security or remediation of vulnerabilities. The number of discovered vulnerabilities increases every year requiring automation of their processing and classification. Assigning an incident automatically to an element of a classification scheme (taxonomy) will improve the response time and reduce the harmful consequences of the incident. To address these challenges, we propose a compositional ML method called CoMLCSIRT for finding adequate reflection of the classification status of an incident or vulnerability. Our model supports Triaging phase in handling security incidents, outperforms the baseline classification models and performs as well as classification made by human experts but in real-time and without delay.