|
| Work | Field | Scenario | Goal | Application of NLP | Complement techniques |
|
| Tamura et al. [13] | Cybersecurity | Industrial control networks | Detect anomalies in packet flow | Similarity model | Markov chain model |
|
| Chambers et al. [14] | Cybersecurity | Twitter | Detect cyberattacks and analyze user behavior | Continuous bag-of-words (CBOW) model and topic-based model (PLDA) | |
|
| Khandpur et al. [15] | Cybersecurity | Twitter | Detect cyberattacks in social media | Similarity model, domain generation algorithm, and dynamic query expansion | |
|
| Ritter et al. [16] | Cybersecurity | Twitter | Detect cyberattacks in social media | Name-entity recognition | Expectation regularization |
|
| Kong et al. [17] | Cybersecurity | Google Play | Evaluate the security of Android apps through user reviews | Bag-of-words (BOW) + classifier (sparse SVM) | Crowdsourcing techniques |
|
| Liao et al. [18] | Cybersecurity | Technical articles | Discovery indicators of compromise | Dependency parsing and topic extraction (POST) | Classifier (SVM), classifier (logistic regression), and graph mining |
|
| Pereira-kohatsu et al. [9] | Hate crime | Twitter | Identify and monitor hate speech | Long short-term memory NN + multilayer perceptron | Classifiers (LDA, QDA, RF, RLR, and SVM) |
|
| Muhammad et al. [19] | Hate crime | Twitter | Classify messages as hate speech, offensive, or nonoffensive | Sequential CNN (SCNN) | |
|
| Gambäck and Sikdar [20] | Hate crime | Twitter | Classify tweets as “racist,” “sexist,” “both,” or “non-hate-speech” | Convolutional NN (CNN) | |
|
| Malmasi and Zampieri [21] | Hate crime | Twitter | Annotate tweets with labels “hate,” “offensive,” or “ok” | Linear SVM | |
|
| Qian et al. [22] | Hate crime | Twitter | Analyze real-life extremists and hate groups | Bidirectional LSTM (bi-LSTM) + deep reinforcement learning | |
|
| Araque and Inglesias [7] | Radicalization | Twitter and online newspapers | Categorize radical users | Sentiment analysis and similarity model | Logistic regression and linear SVM |
|
| Nouh et al. [23] | Radicalization | Twitter | Categorize radical tweets | Language model and sentiment analysis | Classifiers (RF, NN, SVM, and KNN) |
|
| Chen [24] | Radicalization | Dark web | Categorize forum postings | Ensemble SVR | Clustering |
|
| RED-Alert [25] | Radicalization | Social media | Monitor social networks in real time | Semantic analysis, lexical analysis, and domain-specific ontologies | Social network analysis and complex event processing |
|
| Iqbal et al. [26] | Cybercrime | Chat logs | Summarize conversations into crime-related topics | Named-entity recognition, semantic analysis, similarity model | Information visualizer |
|
| Pastrana et al. [27] | Cybercrime | Underground forums | Detect cybercrime topics and identify potential victims | Logistic regression and topic extraction | Social network analysis and clustering (K-means) |
|
| Bhalerao et al. [28] | Cybercrime | Underground forums | Analyze posts and replies for the identification of supply chains | Classifiers, (FT, LR, SVM, and XGBoost) | |
|
| Our work | Cybercrime | Twitter | Identify suspect groups | Similarity model and sentiment analysis | Clustering (K-means) and graph mining |
|
