Analysis of Women Safety in Indian Cities Using Self-Attention-Based Discrete Hopfield Neural Network on Tweets

People often bring attention to the lack of women's safety in Indian cities by sharing their experiences on Twitter. Particularly, those posts marked under movements such as MeToo represent true-life situations and how people feel, providing useful data for safety analysis. Even so, discovering patterns worth noting from such unreliable data is extremely tough. The proposed framework introduces a secure deep learning algorithm that can sort social media messages into safe or harmful. The initial stage uses Adjusted Min-Max Scaling, Decimal Scaling and Statistical Column Normalization to improve how the data is handled. A Self-Attention-based Discrete Hopfield Neural Network (SA-DHNN) is used for labeling, supporting both flexible feature communication and memory learning. In addition, the Migrating Walrus Algorithm (MWA) is implemented to improve the accuracy and convergence of the SA-DHNN model. Experimental testing confirmed that proposed model detects 99.9 % of safety-related content in social media narratives which shows its effectiveness.