This is an unedited manuscript accepted for publication and provided as an Article in Press for early access at the author’s request. The article will undergo copyediting, typesetting, and galley proof review before final publication. Please be aware that errors may be identified during production that could affect the content. All legal disclaimers of the journal apply.
Divya Wagh,
Saminabano Shaikh,
- U.G Student, Student, Department of Computer Engineering, Trinity College of Engineering and Research, Maharashtra, India
- Assistant Professor, Department of Computer Engineering, Trinity College of Engineering and Research, Maharashtra, India
Abstract
This research introduces a machine learning-based approach for examining and assessing women’s safety concerns across social media platforms. The proposed framework aims to identify potentially harmful online interactions, recognize patterns associated with harassment, and support systematic safety assessment. Rather than replacing human judgment with fully automated moderation, the framework serves as an analytical support tool to assist in identifying risks and improving awareness of online threats. The approach focuses on supporting the analysis of online safety concerns and facilitating informed decision-making to promote a safer digital environment for women. It combines natural language processing, image analysis, account and interaction features, temporal patterns, and human review to identify potentially harmful content and prioritize cases by risk. The proposed workflow includes privacy-aware data intake, preprocessing, multimodal feature extraction, ensemble classification, contextual risk scoring, explanation, human validation, and proportionate safety actions. The review also examines challenges related to ambiguous language, cultural context, sarcasm, multilingual content, dataset bias, false alarms, privacy, and cross-platform generalization. Attention is given to responsible AI controls, including data minimization, audit logs, threshold calibration, appeal mechanisms, and continuous evaluation. The framework is aligned with electrical, electronics, and computing environments through mobile devices, edge processing, secure communication gateways, cloud services, and monitoring dashboards. The paper concludes that machine learning can improve the speed and consistency of social-media safety monitoring when it supports, rather than replaces, human judgment and platform governance. In addition, the framework highlights the need to assess the performance of the models by their respective metrics – accuracy, precision, recall, F1 score, confusion matrices – and review the classification errors. A staged approach is suggested starting from a baseline text-classification model and moving to multimodal analysis and to human- supervised monitoring. The framework also emphasizes the importance of ongoing evaluation of model reliability, privacy safeguards and evolving trends in online abuse to enable responsible and effective monitoring of safety.
Keywords: Machine Learning, Online Misogyny, Natural Language Processing, Multimodal AI, Responsible AI
References
- Bailey J, Mathen C. Technology-facilitated violence against women & girls: Assessing the Canadian criminal law response. Can. B. Rev.. 2019;97:664.
- Bardall G. An overview of policy responses and solutions to violence against women in politics. European journal of politics and gender. 2020 Jun;3(2):299-301.
- Fortuna P, Nunes S. A survey on automatic detection of hate speech in text. Acm Computing Surveys (Csur). 2018 Jul 31;51(4):1-30.
- Basile V, Bosco C, Fersini E, Nozza D, Patti V, Pardo FM, Rosso P, Sanguinetti M. Semeval-2019 task 5: Multilingual detection of hate speech against immigrants and women in twitter. InProceedings of the 13th international workshop on semantic evaluation 2019 Jun (pp. 54-63).
- Guest E, Vidgen B, Mittos A, Sastry N, Tyson G, Margetts H. An expert annotated dataset for the detection of online misogyny. InProceedings of the 16th conference of the European chapter of the association for computational linguistics: main volume 2021 Apr (pp. 1336-1350).
- Zeinert P, Inie N, Derczynski L. Annotating online misogyny. InProceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) 2021 Aug (pp. 3181-3197).
- Fersini E, Gasparini F, Rizzi G, Saibene A, Chulvi B, Rosso P, Lees A, Sorensen J. SemEval-2022 task 5: Multimedia automatic misogyny identification. InProceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022) 2022 Jul (pp. 533-549).
- Vidgen B, Harris A, Nguyen D, Tromble R, Hale SA, Margetts H. Challenges and frontiers in abusive content detection. InProceedings of the third workshop on abusive language online 2019 Aug (pp. 80-93).
- Nejadgholi I, Kiritchenko S. On cross-dataset generalization in automatic detection of online abuse. InProceedings of the fourth workshop on online abuse and harms 2020 Nov (pp. 173-183).
- Cascione A, Cerulli A, Manerba MM, Passaro L. Women’s professions and targeted misogyny online. InProceedings of the Tenth Italian Conference on Computational Linguistics (CLiC-it 2024) 2024 Dec (pp. 182-189).
- Arora I, Guo J, Levitan SI, McGregor S, Hirschberg J. A novel methodology for developing automatic harassment classifiers for Twitter. InProceedings of the fourth workshop on online abuse and harms 2020 Nov (pp. 7-15).
- AI N. Artificial intelligence risk management framework (AI RMF 1.0). URL: https://nvlpubs. nist. gov/nistpubs/ai/nist. ai. 2023 Jan;1(0):0-1.
Current Trends in Signal Processing
| Volume | 16 | |
| 03 | ||
| Received | 12/10/2026 | |
| Accepted | 15/10/2026 | |
| Published | 21/10/2026 | |
| Publication Time | 9 Days |