Detecting Sensitive Information from Unstructured Text in a Data-Constrained Environment
Saurabh Anand, Manish Shukla, Sachin Lodha · 2023
For an enterprise, it is important to handle sensitive customer data properly because any data breach or violation can lead to hefty penalties. Past work has looked at various techniques for detecting sensitive data in free-flowing text for a given regulation. However, most of them either produce many false positives or are very specific to certain types of data, for example, email, account number, or social security number. Moreover, machine learning-based methods are difficult to use as finding large amounts of labeled data for training a supervised model poses a serious challenge. In this work, we aim to address the issue of sensitive data discovery in a data-constrained environment by utilizing pre-trained models. We compare their effectiveness in the financial and health domains. Further, we improve the performance of pre-trained models by employing morphological-level features and propose a hybrid model architecture. Our experimental results show that pre-trained models in a data-constrained environment can reduce the turnaround time for sensitive data discovery, thus saving money and effort.