In urban environments, harassment and violence remain significant issues for women, exacerbated by bullying and abusive content prevalent on online social networks (OSNs). This highlights the urgent need for effective measures to assess and enhance women's safety in these digital spaces. Traditional approaches have often fallen short in providing comprehensive safety analysis. This study introduces the Women Safety Prediction using Decision Tree (WSP-DT) classifier, aimed at improving safety predictions in OSN environments. We utilize a Twitter dataset as the foundation for our system, which undergoes preprocessing to eliminate missing values and irrelevant symbols. The Natural Language Toolkit (NLTK) is employed for tokenization, conversion to lowercase, stop-word removal, stemming, and lemmatization of tweets. Subsequently, a text blob protocol is developed to analyze sentiments, categorizing tweets as positive, negative, or neutral. We apply the Term Frequency- Inverse Document Frequency (TF-IDF) method to extract data features based on word and character frequency. Finally, the Decision Tree classifier is employed to discern between fake and genuine tweets through multi-level training. Simulations conducted on the Twitter dataset demonstrate that the proposed WSP-DT classifier outperforms existing methods, showcasing its potential for enhancing women's safety in online environments.

Women safety, online social network, Natural language toolbox kit.