We introduce CyBERT, a cybersecurity feature claims classifier based on bidirectional encoder representations from transformers and a key component in our semi-automated cybersecurity vetting for industrial control systems (ICS). To train CyBERT, we created a corpus of labeled sequences from ICS device documentation collected across a wide range of vendors and devices. This corpus provides the foundation for fine-tuning BERT’s language model, including a prediction-guided relabeling process. We propose an approach to obtain optimal hyperparameters, including the learning rate, the number of dense layers, and their configuration, to increase the accuracy of our classifier. Fine-tuning all hyperparameters of the resulting model led to an increase in classification accuracy from 76% obtained with BertForSequenceClassification’s original architecture to 94.4% obtained with CyBERT. Furthermore, we evaluated CyBERT for the impact of randomness in the initialization, training, and data-sampling phases. CyBERT demonstrated a standard deviation of ±0.6% during validation across 100 random seed values. Finally, we also compared the performance of CyBERT to other well-established language models including GPT2, ULMFiT, and ELMo, as well as neural network models such as CNN, LSTM, and BiLSTM. The results showed that CyBERT outperforms these models on the validation accuracy and the F1 score, validating CyBERT’s robustness and accuracy as a cybersecurity feature claims classifier.

There is an urgent need in many critical infrastructure sectors, including the energy sector, for attaining detailed insights into cybersecurity features and compliance with cybersecurity requirements related to their Operational Technology (OT) deployments. Frequent feature changes of OT devices interfere with this need, posing a great risk to customers. One effective way to address this challenge is via a semi-automated cyber-physical security assurance approach, which enables verification and validation of the OT device cybersecurity claims against actual capabilities, both pre- and post-deployment. To realize this approach, this article presents new methodology and algorithms to automatically identify cybersecurity-related claims expressed in natural language form in ICS device documents. We developed an identification process that employs natural language processing (NLP) techniques with the goal of semi-automated vetting of detected claims against their device implementation. We also present our novel NLP components for verifying feature claims against relevant cybersecurity requirements. The verification pipeline includes components such as automated vendor identification, device document curation, feature claim identification utilizing sentiment analysis for conflict resolution, and reporting of features that are claimed to be supported or indicated as unsupported. Our novel matching engine represents the first automated information system available in the cybersecurity domain that directly aids the generation of ICS compliance reports.

This paper presents our research approach and findings towards maximizing the accuracy of our classifier of feature claims for cybersecurity literature analytics, and introduces the resulting model ClaimsBERT. Its architecture, after extensive evaluations of different approaches, introduces a feature map concatenated with a Bidirectional Encoder Representation from Transformers (BERT) model. We discuss deployment of this new concept and the research insights that resulted in the selection of Convolution Neural Networks for its feature mapping aspects. We also present our results showing ClaimsBERT to outperform all other evaluated approaches. This new claims classifier represents an essential processing stage within our vetting framework aiming to improve the cybersecurity of industrial control systems (ICS). Furthermore, in order to maximize the accuracy of our new ClaimsBERT classifier, we propose an approach for optimal architecture selection and determination of optimized hyperparameters, in particular the best learning rate, number of convolutions, filter sizes, activation function, the number of dense layers, as well as the number of neurons and the drop-out rate for each layer. Fine-tuning these hyperparameters within our model led to an increase in classification accuracy from 76% obtained with BertForSequenceClassification’s original model to a 97% accuracy obtained with ClaimsBERT.