Predators often use the dark web to discuss and share Child Sexual Abuse Material (CSAM) because the dark web provides a degree of anonymity, making it more dificult for law enforcement to track the criminals involved. In most countries, CSAM is considered as forensic evidence of a crime in progress. Processing, identifying and investigating CSAM is often done manually. This is a time-consuming and emotionally challenging task. In this paper, we propose a novel model based on artificial intelligence algorithms to automatically detect CSA text messages in dark web forums. Our algorithms have achieved impressive results in detecting CSAM in dark web, with a recall rate of 89%, a precision rate of 92.3% and an accuracy rate of 87.6%. Moreover, the algorithms can predict the classification of a post in just 1 microsecond and 0.3 milliseconds on standard laptop capabilities. This makes it possible to integrate our model into social network sites or edge devices to for real-time CSAM detection.
avoid detection and hide their activities.
In this context, we propose a CSAM detection intelliIn general, Child Sexual Abuse Material (CSAM) includes gence model based on both classical Machine Learning any visual, written or audio material that depicts or de- (ML) and Deep Learning (DL) techniques. Our CSAM scribes sexual abuse of children. This can include pho- detection model can be used to monitor and remove CSA tographs, videos, stories, chats, comments, drawings or texts on online platforms in real-time and with high acany other media1. The production and distribution of curacy, providing better protection for children. We have CSAM has negative impacts on victims and society. Vic- also created a manually labelled dataset of CSAM and tims can live with long psychological, emotional, and non-CSAM content that can be used to train and test physical harm [1]. A high volume of CSAM is created CSAM detection algorithms. In the future, our model will and shared daily on both surface web platforms such as be able to detect perpetrator behaviours, collect forensic social network sites and dark web forums. It is not vi- evidence, and extract valuable knowledge for child agenable for human experts to investigate, detect and prevent cies, hotlines, education programs and policy makers. CSAM manually [2]. However, automatically detecting The remainder of the paper is organised as follows. and analysing online CSA text can be extremely chal- In the Section 2 we review the related work. Section lenging and time-consuming, due to language complex- 3 presents our system architecture and the machine ity, contextual ambiguity, dynamic nature of language learning and deep learning algorithms. The evaluation and large volume of data. This is particularly the case methodology and experimental results for the system are for CSAM shared on the dark web, where privacy and shown in Section 4. Finally, we conclude and give some anonymity are prioritized. Moreover, perpetrators often future directions in Section 5. use code words, slang, or other forms of obfuscation to
and DL models to process CSA text. In [8], Naive Bayes
(NB), Support Vector Machine (SVM) and Random Forest
(RF) were applied to detect online abusive and bullying
comments on Facebook and Twitter. In [9], the histogram 3.1. System Architecture
gradient boosted decision trees were exploited for
predatory chat conversation detection. In [
Dark web data was also processed in [13], [14], [15], sification. In that, the Supervised Method component [16] and [17]. However, the approaches did not auto- implements NB, SVM, LSM and BERT algorithms. The matically detect CSA text on the dark web by using post algorithms tokenizes the post texts and transforms them contents and artificial intelligence. In [ 13], the 450 au- into vector representation using TF.IDF2 (in NB and SVM) thorised hidden service sites were manually classified. or embedding layers3 (in LSTM and BERT). In [14], the authors analysed seven popular dark web The Evaluation component is used to determine the sites to monitor the sites by using their metadata, e.g. the execution times (i.e., training time and prediction time) number of users, site names and common users in sites. and the classification performance metrics (i.e., precision, In [15], the authors statistically analysed some simple recall and accuracy) of each combination of algorithms. metadata e.g. victim ages and the number of CSAM re- To avoid overfitting, we apply 5-fold cross-validation of ports per year. In [16], K-Means algorithm was applied our dataset. Then the algorithm uses 4 folds for the trainto cluster the forum comments into the selected seven ing set and the remaining fold for the testing set. This labels, i.e. breach, financial, drug, vendor, account, prod- process is repeated until every fold serves as the testing uct and other. In [17], the authors manually analysed transcripts of 53 anonymous suspects in United Kingdom to understand suspects’ interaction behaviors and sexual interests. 2https://scikit-learn.org/stable/modules/generated/sklearn.featur e_extraction.text.TfidfVectorizer.html 3https://www.tensorflow.org/api_docs/python/tf/keras/layers/Em bedding set. The average of the recorded classification measures of five rounds are the classification performance measures for the algorithm. Finally, we analyse experimental results on the data set to recommend the best algorithm for CSAM text classification in the dark web.
NB is a specific type of probabilistic classifier that relies on applying Bayes’ theorem with certain simplifying assumptions. NB is widely used in natural language processing, spam filtering, and other applications where it is necessary to classify items into diferent categories based Figure 2: sexual abuse single words in dark web forums on probabilistic features. It assumes that the features are strongly independent to simplify computation. We used the Gaussian Naive Bayes algorithm implemented in [20], with parameters: ℎ= 1 and _= True.
Where, ℎ is the additive smoothing parameter and _ determines whether to learn class’ prior probabilities or not.
SVM represents patterns as points in space and divides the data points by a clear gap. It constructs a maximum margin separator and can perform a non-linear classification by using the so-called kernel trick. We used the C-support vector classification algorithm implemented in [21], with parameters: = 1.0, = ’linear’, Figure 3: sexual abuse two-word-phrases in dark web forums = 3 and = ’auto’. Where, is the regularization parameter. is the used kernel type. is the degree of the polynomial kernel function 3.3. Training and Testing Datasets and is kernel coeficient.
LSTM is a special kind of Recurrent Neural Network Our first step is to create a labelled dataset that can be (RNN). RNN is a type of neural network commonly used for training or fine-tuning our classifier. The laused to develop natural language processing models. belled dataset used for our study was collected and supRNN remembers the sequence of the data and exploits plied by the company Web-IQ, which provided us with data patterns and feedback loops for prediction. LSTM over 352,000 posts from 8 dark web forums in 2022, of was applied to avoid the long-term dependency which approximately 221,000 were in English. problem in regular RNN. We used the Bidirectional- Using a dictionary of 12,628 Sexual Abuse Phrases LSTM algorithm implemented in [22], with param- (SAPs) extracted from THORN project4 and Web-IQ eters: = (1000, 128, input_length=200), dark web forums5, we were able to detect approximately (LSTM(64)), (0.5) and 177,000 English posts with no SAP and approximately (1, activation=’sigmoid’) 44,000 English posts with at least one SAP. This provides
BERT is a language model using the transformer en- us with a high level grouping of posts, but with refinecoder architecture to process tokens in text. BERT ap- ment required to allow for CSAM posts that does not plies pre-training and fine-tuning. Pre-training is an contain any SAPs, and vice versa. Figures 2 and 3 show unsupervised way on a general large corpus of text to the word clouds of single words and two-word-phrases create BERT model. Fine-tuning is a supervised train- related to sexual abuse, extracted from post contents in ing BERT model on a specific downstream task with dark web forums. The size of each word in the clouds relatively few labels, because the general linguistic pat- represents its frequency in the forums. terns have already been learnt during pre-training. We From the group of 177,000 posts with no SAP, experts used BERT algorithm implemented in [23], with pa- randomly selected 2,000 non-CSAM posts and 500 CSAM rameters: (bert_en_uncased_preprocess_3, posts. From the group of 44,000 posts with at least one bert_en_uncased_L-12_H-768_A-12_4), (1, activa- SAP, experts randomly selected 2,000 CSAM posts and tion=’sigmoid’) and =’adam’. 100 non-CSAM posts. Ultimately, our manually labelled
5https://web-iq.com/solutions/osint-on-premises dataset contains 4,600 posts from the dark web, including 2,500 CSAM posts and 2,100 non-CSAM posts.
scikit-learn library 1.2.2 (for NB and SVM), keras library 1.1.2 run on top of tensorflow library 2.10.0 (for LSTM and BERT). All experiments were run under Windows 10 (64-bit) on a Dell laptop with an Intel Core i7 CPU (3.00 GHz) and 16 GB memory.
For the purpose of measuring the quality of the predicted classes of posts compared to the correct classes, we apply the most commonly used metrics namely accuracy, precision and recall ([24, 25]). The metrics are derived from four categories in the confusion matrix: True Positive (TP), False Positive (FP), True Negative (TN) and False Negative (FN) as follows: • TP: Posts in which the algorithm predicted CSAM and the correct class was also CSAM. • FP: Posts in which the algorithm predicted CSAM, but the correct class was non-CSAM. • TN: Posts in which the algorithm predicted non
CSAM and the correct class was non-CSAM. • FN: Posts in which the algorithm predicted non
CSAM, but the correct class was CSAM.
ratio between the correctly classified samples to the total + number of samples: = + + + . The accuracy puts the same emphasis on all these factors. However, when categorising pairs, there is usually a bias: it is much easier to identify true negatives correctly, due to their large number. So, we also look at the the precision = + and the recall = + .
Using the 5-fold cross-validation methodology, each experimental round includes a training set of 3,680 posts (2,000 CSAM and 1,680 non-CSAM) and a testing set of 920 posts (500 CSAM and 420 non-CSAM). Table 1 presents the average training time, average prediction time, average precision, average recall and average accuracy of four algorithm combinations as follows: • NB: The training time and prediction time were 0.5 and 0.001 seconds, respectively. The precision We proposed and implemented a novel algorithm based was 76.1%, recall was 89% and accuracy was 78.8%. on machine learning and natural language processing to automatically detect and classify CSAM text post content in dark web. In the experimental evaluation on the
The combination of the NB algorithm has the fastest execution time, taking only about 1 microsecond to detect a post on our laptop’s capabilities. The second best performing algorithm is SVM, which takes about 0.3 milliseconds. These fast prediction times make our models well-suited for processing CSA text in real-time on social networks. Additionally, our models can run on edge devices with limited computational resources and power supply.
In terms of classification precision, the SVM combination performs the best with 92.3%, followed by LSTM and BERT as the second and third-best performers, respectively. Meanwhile, the NB combination has the highest recall rate of 89%, followed by LSTM as the second-best performer. When it comes to accuracy, SVM is the best with 87.6% which is slightly higher than LSTM with 87.1%.
The BERT algorithm has long training and prediction times, and it is not suitable for binary classification of CSAM posts in dark web.
part of the N-Light project which is funded by the Safe Online Initiative of End Violence and the Tech Coalition through the Tech Coalition Safe Online Research Fund (Grant number: 21-EVAC-0008-Technological University Dublin). Dr. Vuong Ngo has conducted the research while serving as a data scientist at TU Dublin. dataset of 4,600 CSAM and non-CSAM posts with 5-fold cross-validation, the combination of NB algorithm performed the best in terms of classification recall and execution time. On the other hand, the SVM combination performed the best in terms of classification precision and accuracy, and was the second-best in execution time. The choice of NB and SVM depends on the specific goals and requirements of the CSAM classification task. NB is maximize the number of true positives which could be useful identifying and removing CSAM posts from online platforms to protect potential victims. On the other hand, SVM is minimize false positives which could be useful for identifying CSAM posts to extract information about potential predators and victims for investigative purposes.
As part of our future work, functional APIs will be implemented to create a user-friendly web application. Furthermore, we aim to leverage the metadata associated with CSAM posts to identify the characteristics, conversation and behaviours of perpetrators. This information can be valuable in developing more efective models for preventing and addressing CSA text on social media platforms. We also will recognise named entities in CSA text to supply important concepts for ML models [26]. M. Liechti, V. Lenders, Blackwidow: Monitoring the dark web for cyber security information, in: the 11th International Conference on Cyber Conflict (CyCon), volume 900, 2019, pp. 1–21. doi:10.239 19/CYCON.2019.8756845. [15] E. Kokolaki, E. Daskalaki, K. Psaroudaki, M. Christodoulaki, P. Fragopoulou, Investigating the dynamics of illegal online activity: The power of reporting, dark web, and related legislation, Computer Law & Security Review 38 (2020) 105440. doi:10.1016/j.clsr.2020.105440. [16] S. Nazah, S. Huda, J. H. Abawajy, M. M. Hassan, An unsupervised model for identifying and characterizing dark web forums, IEEE Access 9 (2021) 112871– 112892. doi:10.1109/ACCESS.2021.3103319. [17] J. Woodhams, J. A. Kloess, B. Jose, C. E. HamiltonGiachritsis, Characteristics and behaviors of anonymous users of dark web platforms suspected of child sexual ofenses, Frontiers in Psychology 12 (2021) 623668. doi:10.3389/fpsyg.2021.623668. [18] T. Tran, L. Nguyen, V. Ngo, Machine learning based english sentiment analysis, Journal of Science and Technology 52(4D) (2014) 142–155. doi:https:// doi.org/10.48550/arXiv.1905.06643. [19] S. J. Russell, P. Norvig, Artificial Intelligence: A
Modern Approach (4th Edition), Pearson, 2022. [20] Scikit-learn, Multinomial naive bayes, https://scik it-learn.org/stable/modules/generated/sklearn.na ive_bayes.MultinomialNB.html#sklearn.naive_b ayes.MultinomialNB, 2023. Version 1.2.2, accessed April 01, 2023. [21] Scikit-learn, C-support vector classification, https: //scikit-learn.org/stable/modules/generated/sklear n.svm.SVC.html, 2023. Version 1.2.2, accessed April 01, 2023. [22] Keras, Long short term memory, https://www.te nsorflow.org/api_docs/python/tf/keras/layers/LS TM, 2023. Version 1.1.2, accessed April 01, 2023. [23] Keras, Text classsification with bert, https://www. tensorflow.org/text/tutorials/classify_text_with_ bert, 2023. Version 1.1.2, accessed April 01, 2023. [24] V. M. Ngo, T. V. T. Duong, T. B. T. Nguyen, P. T.
Nguyen, O. Conlan, An eficient classification algorithm for traditional textile patterns from diferent cultures based on structures, Journal on Computing and Cultural Heritage (JOCCH) 14(4) (2021) 1–22. doi:https://doi.org/10.1145/3465381. [25] A. Tharwat, Classification assessment methods, Applied Computing and Informatics 17(1) (2021) 168–192. doi:https://doi.org/10.1016/j.ac i.2018.08.003. [26] V. M. Ngo, G. Munnelly, F. Orlandi, P. Crooks, D. O’Sullivan, O. Conlan, A semantic search engine for historical handwritten document images, in: G. Berget, M. M. Hall, D. Brenn, S. Kumpulainen