We congratulate Dr. Mengyuan Zhang and Dr. Lianying Zhao on their recent publication.

One of VU’s Network institute Research Visits has produced a paper published at the 2026 CHI Conference on Human Factors in Computing Systems (presented in Barcelona Spain in April, 2026).

The work aims to examine how NLP-based assistive tools impact cybersecurity professionals and amateurs when evaluating software vulnerabilities. Currently, manual effort is involved to assign a score to reported vulnerabilities (security issues/bugs) to reflect their severity for decision makers, which process has been well studied and found to be error-prone, time-consuming and potentially inconsistent. A user study with 389 selected online participants was conducted based on an existing NLP-based tool VIET, and a new Web-based UI with hints and visual guides to help each analyst with the scoring. The results show that the tool does not consistently improve accuracy/efficiency but instead can help for certain vulnerability types, metrics, demographic groups, and can increase user confidence. More interestingly, the tool can support training for manual assessment tasks that are hard to automate thanks to the learning effect.

During the visit, Dr. Mengyuan Zhang (the host) and Dr. Lianying Zhao from Carleton University (the visiting scholar) worked with Siqi and Minjie on designing and conducting the large-scale online survey via Amazon MTurk and Prolific, followed by the collection and analysis of the survey data with numerous discussions, where the feedback from other VU colleagues and students was invaluable.

Are you interested in collaborating with researchers from abroad? Take a look at our Research Visits program – the call is open until June 15.