Abstract
This research paper comes up with an approach that attempts to enhance user password guessing effectiveness through the analysis of publicly accessible social media data. The key emphasis is the extraction of personalized information that can be utilized during password creation. For the sake of this paper, a system was put in place to examine images shared on Twitter with the aim of recognizing celebrities, food, places, and web memes. In addition, user profiles on LinkedIn were mined to obtain data related to work experience, skills, educational history, and first and last names. Data collected were utilized to derive probabilistic models for passwords based on Probabilistic Context-Free Grammar (PCFG). The proposed method depicts the ability of increased password prediction accuracy through tailoring the model to the specific traits of users, which may be applicable both in cybersecurity fields and investigations relative to social engineering.