Open research questions in Spam and Phishing Detection
29 unresolved questions extracted from the limitations and future-work sections of 334 Spam and Phishing Detection papers in our library. Each links back to the study that raised it.
What the literature leaves open
The key limitations of existing unimodal and hybrid detection systems are overcome by the integration of URL risk analysis, visual QR code integrity inspection, and conditional anomaly detection layers.
Future research may explore transformer-based sentiment models such as BERT or RoBERTa, which will fuse user behaviour with temporal elements to create improved fraud detection systems, while developing MCDM systems that enable stakeholders to modify criterion weightings in real time.
Enhancing E-Commerce Trust: Integrating Fraud Detection with Sentiment Analysis and MCDM · 2026 · DOIThe proposed deep learning-based approach contributes to the development of automated cybersecurity systems that can assist in identifying malicious websites and reducing the risks associated with phishing attacks, while future research may focus on incorporating larger datasets, exploring hybrid deep learning architectures, and developing real time phishing detection systems for practical cybersecurity applications.
Deep Learning-Based Phishing URL Detection Using Deep Neural Network and Convolutional Neural Networks · 2026 · DOI• Incorporate machine learning for detecting phishing attacks • Create an application for use on mobile phones • Include real-time URL analysis capability • Upgrade detection algorithms to counter new methods of phishing attacks • Increase accuracy and performance REFERENCES J. Ma, L. K. Saul, S. Savage, and G. M. Voelker, “Beyond blacklists: Learning to detect malicious web sites from suspicious URLs,” Proceedings of the ACM SIGKDD, pp. 1245–1254, 2009. A. Le, A. Markopoulou, and M. Faloutsos, “PhishDef: URL names say it all,” IEEE INFOCOM, pp. 191–195, 2011. M. Khonji, Y. Iraqi, and A. Jones, “Phishing detection: A literature survey,” IEEE Communications Surveys & Tutorials, vol. 15, no. 4, pp. 2091–2121, 2013. R. Verma and N. Hossain, “Semantic feature selection for text with application to phishing email detection,” IEEE Conference on Communications and Network Security, pp. 455–463, 2014. S. Garera, N. Provos, M. Chew, and A. D. Rubin, “A framework for detection and measurement of phishing attacks,” ACM Workshop on Recurring Malcode, pp. 1–8, 2007. N. Abdelhamid, A. Ayesh, and F. Thabtah, “Phishing detection based on associative classification data mining,” Expert Systems with Applications, vol. 41, no. 13, pp. 5948–5959, 2014. S. Rao and K. Kumar, “Phishing website detection using URL-based feature extraction and classification,” International Journal of Advanced Computer Science and Applications, vol. 11, no. 5, pp. 210–216, 2020. P. Prakash, M. Kumar, R. R. Kompella, and M. Gupta, “PhishNet: Predictive blacklisting to detect phishing attacks,” IEEE INFOCOM, pp. 1–5, 2010. A. K. Jain and B. B. Gupta, “Phishing detection: Analysis of visual similarity based approaches,” Security and Communication Networks, vol. 2017, pp. 1–20, 2017. S. Marchal, J. Francois, R. State, and T. Engel, “PhishStorm: Detecting phishing with streaming analytics,” IEEE Transactions on Network and Service Management, vol. 11, no. 4, pp. 458–471, 2014. VII. CONCLUSION In terms of the project PhishAlert – A Risk-Based URL Detection System, we have a good approach that will help you find phishing links effectively through rule-based analysis and verification via the external website.
Federated learning approaches for phishing detection introduce accuracy-computational cost trade-offs when privacy-preserving techniques (differential privacy, homomorphic encryption) are applied, but the quantitative relationship between privacy budgets and detection performance degradation in federated phishing detection systems remains inadequately characterized.
Phishing in the age of distributed intelligence: taxonomies, detection strategies, and the emerging role of federated learning · 2026 · DOIWhile federated learning has been applied to URL inspection, email classification, and vishing detection, the effectiveness of FL across different phishing detection domains with varying data distributions and model architectures (LSTM, CNN-LSTM, MLP, FFNN) has not been comprehensively compared under standardized evaluation conditions.
Phishing in the age of distributed intelligence: taxonomies, detection strategies, and the emerging role of federated learning · 2026 · DOIData heterogeneity and non-IID distributions cause increased divergence, slower convergence, and lower global model accuracy in federated learning-based phishing detection systems. Specific techniques to enhance FL resilience under non-IID phishing data across clients require further development and empirical validation.
Phishing in the age of distributed intelligence: taxonomies, detection strategies, and the emerging role of federated learning · 2026 · DOIExisting phishing detection datasets have remained largely static and unchanged for several years, with most being centralized rather than distributed. These datasets do not reflect the non-IID distributional characteristics and data heterogeneity present in real federated learning environments, limiting the ability to properly evaluate federated learning approaches for phishing detection.
Phishing in the age of distributed intelligence: taxonomies, detection strategies, and the emerging role of federated learning · 2026 · DOIAlthough the proposed system demonstrates promising results in detecting fake reviews, there are several opportunities for further improvements and enhancements. One potential direction for future work is the integration of advanced deep learning models such as Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM), and Transformer-based architectures. These models can capture deeper semantic relationships and contextual dependencies in review text, which may further improve detection accuracy. Another improvement could involve the incorporation of aspect-based sentiment analysis to analyze specific product features mentioned in reviews, such as price, quality, and delivery. This approach can provide more detailed insights into review authenticity by examining how users express opinions about different product aspects. Future systems may also integrate user behavioral analysis, including reviewer activity patterns, posting frequency, and rating behavior. Combining textual analysis with behavioral data can enhance the ability to detect suspicious review patterns. Additionally, the system can be extended to support multilingual review detection, allowing the detection of fake reviews written in different languages. This feature would make the system more applicable to global e-commerce platforms. Another possible improvement is the deployment of the model in large-scale cloud environments, enabling real-time analysis of massive volumes of reviews generated across multiple online marketplaces. Overall, future research can focus on combining advanced machine learning techniques, behavioral analysis, and large-scale deployment to build more intelligent and robust fake review detection systems.
The web-based implementation uses Flask/Streamlit, but the paper does not address how the framework integrates with existing security infrastructure (email gateways, browser extensions, corporate proxies) or handles browser-based QR scanning with varying camera quality and lighting conditions.
While ethical considerations address data protection and transparency, the paper does not evaluate adversarial robustness: how the model performs against evasion attacks (e.g., homograph attacks, obfuscated URLs, dynamically generated QR codes with steganographic encoding).
The paper states datasets were balanced to minimize bias, but does not specify the class distribution (ratio of legitimate to phishing URLs/QR codes), sampling strategy, or whether the balanced dataset reflects real-world phishing prevalence, which affects practical false positive rate generalization.
The framework comparison with baseline methods (blacklist-based systems, manual verification) lacks quantitative performance metrics such as detection time comparison, false negative rates, or computational resource consumption (CPU/memory) between the proposed ML-based approach and traditional rule-based systems.
The system demonstrated 100+ hours continuous stability without crashes, but no evaluation of performance degradation under high-volume concurrent requests (e.g., scanning 1000+ URLs/QR codes simultaneously) or with emerging phishing techniques post-deployment is documented.
Feature engineering for URL analysis includes phishing score, suspicious keyword frequency, and redirection count, but the paper does not specify which machine learning model architecture (Random Forest, SVM, Neural Network) was ultimately selected or how PCA dimensionality reduction impacts detection accuracy for QR codes versus URLs separately.
The framework was evaluated on labeled phishing datasets and real-time inputs, but the paper does not disclose which specific datasets were used (size, source, temporal range, phishing variants covered). Cross-dataset consistency is mentioned as tested, but the actual datasets and their characteristics are not identified, preventing reproducibility and comparative analysis.
The QR code threat detection accuracy is reported at ~90%, which is lower than URL phishing detection accuracy (92-97%). The paper does not specify which types of malicious QR code patterns (encoded payloads, obfuscation techniques, redirect mechanisms) are responsible for the 10% detection failures, limiting understanding of QR-specific vulnerability exploits.
KNN's performance degradation in text classification is due to high dimensionality of textual data, which increases computational cost and makes distance-based calculations less meaningful.
The findings reported in this paper, other than being supported by a thorough experimental methodology and interesting on their own, also pave the way for further investigation on the novel issue of fake Twitter followers.
Most-cited papers in Spam and Phishing Detection
- Fame for sale: Efficient detection of fake Twitter followers · Decision Support Systems · 2015 · 374 citations
- Fake online reviews: Literature review, synthesis, and directions for future research · Decision Support Systems · 2020 · 293 citations
- Applying machine learning and natural language processing to detect phishing email · Computers & Security · 2021 · 152 citations
- Susceptibility to phishing on social network sites: A personality information processing model · Computers & Security · 2020 · 132 citations
- Mitigation strategies against the phishing attacks: A systematic literature review · Computers & Security · 2023 · 119 citations
- Elevated Learning based Secured Phishing Website Identification Methodology using Artificial Intelligence Assistance · 2024 · 117 citations
- PhiUSIIL: A diverse security profile empowered phishing URL detection framework based on similarity index and incremental learning · Computers & Security · 2023 · 107 citations
- Prompt Engineering or Fine-Tuning? A Case Study on Phishing Detection with Large Language Models · Machine Learning and Knowledge Extraction · 2024 · 105 citations
- A one-class classification approach for bot detection on Twitter · Computers & Security · 2020 · 101 citations
- From conflicts and confusion to doubts: Examining review inconsistency for fake review detection · Decision Support Systems · 2021 · 95 citations
Most recent work
- An Efficient Feature Selection Technique to Enhance Spam Email Detection · TEM Journal · 2026
- DETECTION OF INVALID CLICKS IN DIGITAL ADVERTISING USING ARTIFICIAL INTELLIGENCE TECHNIQUES · JDEBM · 2026
- Comparative Study of Machine Learning Algorithms for E-mail Spam Detection · International Journal of Mathematics And Computer Research · 2026
- Real-Time Phishing URL Detection Pipeline in the PHISHRADOR System · International Scientific Journal of Engineering and Management · 2026
- Deep learning-based phishing classification framework for accurate detection using optimized URL intelligence · Scientific Reports · 2026
- Machine Learning-Based Phishing URL Detection System Using Feature Engineering and Classification Models · International Journal of Creative and Open Research in Engineering and Management · 2026
- Fake Online Product Review Detection Using Supervised and Semi-Supervised Opinion Mining · Research Digest on Engineering Management and Social Innovations · 2026
- Classifiers and Algorithms in Spam Detection · INTERNATIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING AND MANAGEMENT · 2026
- CNN-OOA-Based Cyber Threat Detection: Protecting E-Learning from Phishing · Arabian Journal for Science and Engineering · 2026
- CyberCore Elite: A Cloud-Based Cybersecurity Audit and Threat Intelligence Platform · INTERNATIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING AND MANAGEMENT · 2026
Find a gap in your own Spam and Phishing Detection sub-topic
This page shows what the Spam and Phishing Detection literature already flags as unresolved. To narrow it to your specific question, run the guided finder — it searches the gap library on demand and checks candidates against 250M+ OpenAlex works.
Open the Research Gap Finder →